NXP expands its Scalable Machine Learning portfolio

2 mins read

NXP Semiconductors is enhancing its machine learning development environment and product portfolio.

The company has established an exclusive, strategic partnership with the Canadian-based Au-Zone Technologies to expand its eIQ Machine Learning (ML) software development environment with easy-to-use ML tools and is expanding its offering of silicon-optimised inference engines for Edge ML.

NXP has also announced that it has been working with Arm as the lead technology partner in evolving Arm Ethos-U micro NPU (Neural Processing Unit) architecture to support applications processors. NXP will integrate the Ethos-U65 micro NPU into its next generation of i.MX applications processors to deliver energy-efficient, cost-effective ML solutions for the fast-growing Industrial and IoT Edge.

“NXP’s scalable applications processors deliver an efficient product platform and a broad ecosystem for our customers to quickly deliver innovative systems,” said Ron Martino, senior vice president and general manager of Edge Processing business line at NXP Semiconductors. “Through these partnerships with both Arm and Au-Zone, in addition to technology developments within NXP, our goal is to continuously increase the efficiency of our processors while simultaneously increasing our customers’ productivity and reducing their time to market. NXP’s vision is to help our customers achieve lower cost of ownership, maintain high levels of security with critical data, and to stay safe with enhanced forms of human-machine-interaction.”

Au-Zone’s DeepView ML Tool Suite will augment eIQ with an intuitive, graphical user interface (GUI) and workflow, enabling developers of all experience levels to import datasets and models, rapidly train, and deploy NN models and ML workloads across the NXP Edge processing portfolio.

NXP’s eIQ-DeepView ML Tool Suite will provide developers with advanced features to prune, quantize, validate, and deploy public or proprietary NN models on NXP devices. It’s on-target, graph-level profiling capability is intended to provide developers with unique, run-time insights in order to optimise NN model architectures, system parameters, and run-time performance. By adding Au-Zone’s DeepView run-time inference engine to complement open source inference technologies in NXP eIQ, users will be able to quickly deploy and evaluate ML workloads and performance across NXP devices. A key feature of this run-time inference engine is that it optimises the system memory usage and data movement uniquely for each SoC architecture.

Commenting Brad Scott, CEO of Au-Zone, said, “We created DeepView ML to provide developers with intuitive tools and inferencing technology, so this partnership represents a great union of world class silicon, run-time inference engine technology, and a development environment that will further accelerate the deployment of embedded ML features. This partnership builds on a decade of engineering collaboration with NXP and will serve as a catalyst to deliver more advanced Machine Learning technologies and turnkey solutions as OEM’s continue to transition inferencing to the Edge.”

To accelerate machine learning in a wider range of Edge applications, NXP said that it will expand its popular i.MX applications processors for the Industrial and IoT Edge with the integration of the Arm Ethos-U65 microNPU, complementing the previously announced i.MX 8M Plus applications processor with integrated NPU.

The NXP and Arm technology partnership is focused on defining the system level aspects of this micro NPU which supports up to 1 TOPS (512 parallel multiply accumulate operations at 1GHz). The Ethos-U65 maintains the MCU-class power efficiency of the Ethos-U55 while extending its applicability to higher performance Cortex-A-based system-on-chip (SoC)s. The Ethos-U65 micro NPU works in concert with the Cortex-M core already present in NXP’s i.MX families of heterogeneous SoCs, resulting in improved efficiency.