Project Trillium platform for deep learning accelerator architecture

NVIDIA announced that it has entered into a partnership with Arm to apply deep learning inference capabilities to billions of mobile, consumer electronics, and Internet of Things (IoT) devices targeting the global market.

NVIDIA and Arm announced that through a partnership, they will integrate the open-source NVIDIA Deep Learning Accelerator (NVDLA) architecture into Arm's Project Trillium platform for machine learning. This collaboration will enable IoT chip companies to more easily apply artificial intelligence to their designs and provide affordable intelligent products to billions of consumers worldwide.

Deepu Talla, Vice President and General Manager of Autonomous Machinery at NVIDIA, explained, "Inference tasks will become a core capability for all IoT devices in the future," adding, "Through our partnership with Arm, we will lead this trend by making it easier for hundreds of chip companies to adopt deep learning technology."

Rene Haas, Head of Arm’s IP Group and Senior Vice President, stated, “To realize Arm’s vision of connecting approximately one trillion IoT devices, accelerating artificial intelligence at the edge is paramount,” adding, “By introducing NVDLA to the Arm Project Trillium platform, the overall ecosystem is able to take a step closer to this vision, powered by the expertise and capabilities of both companies in the fields of AI and IoT.”

NVDLA, based on NVIDIA Xavier, the world's most powerful System-on-a-Chip (SoC) for autonomous driving, is an open architecture that promotes a standardized method for designing deep learning inference accelerators. In particular, through its modular architecture, it is designed to be highly scalable, simple to configure, and to simplify integration and portability.

NVDLA offers many advantages for accelerating the application of deep learning inference. NVDLA is supported by NVIDIA's powerful developer tools, including the next version of the programmable deep learning accelerator TensorRT. As an open-source design, the research community and others can regularly add advanced features to edge devices.

The integration of NVDLA and Arm’s Project Trillium will lay the foundation for deep learning developers to apply Arm’s flexibility and scalability to IoT devices across a wide range of fields to achieve peak performance.

Karl Freund, a senior deep learning expert at Moor Insights & Strategy, commented on the partnership between the two companies, stating, "The partnership between the two companies will create a win-win effect for IoT, mobile, and embedded chip companies in designing accelerated AI inference solutions," and added, "It is very reasonable for NVIDIA, an unrivaled leader in AI, and Arm, a leader in IoT, to form a partnership in the IP field."