Mouser to Supply NXP's Ara240 Discrete Neural Processing Unit for AI, Edge, Industrial and Robotics Applications
TECHWORLD ·
✦ AI Summary
Mouser Electronics announced the availability of NXP Semiconductors' Ara240 discrete neural processing unit (DNPU).
The Ara240 DNPU supports real-time generative AI implementation and supports up to 40eTOPS of performance and up to 16GB LPDDR4.
Mouser also supplies the ARA2-M2-16G-GT AI accelerator module based on the Ara240 DNPU, saying it has an M.2 M-Key design, an x4 configuration and support for up to 4 PCIe 4.0 lanes.
Mouser Electronics announced the availability of NXP Semiconductors' Ara240 discrete neural processing unit (DNPU). The Ara240 DNPU supports real-time generative AI implementation.
The Ara240 DNPU also supports industrial automation systems, autonomous robots, smart infrastructure, advanced human-machine interface (HMI) platforms, and a range of edge applications.
Mouser said the Ara240 DNPU combines balanced computing performance with high off-chip bandwidth. On that basis, it supports efficient execution of LLMs and VLMs and delivers low latency, lower operating costs, and enhanced data privacy.
The Ara240 architecture is designed to support advanced multimodal and transformer-based workloads. The Ara240 delivers up to 40eTOPS (equivalent tera-operations per second) of performance.
The Ara240 supports up to 16GB LPDDR4. Mouser said this allows large, complex models to run directly and smoothly at the edge without relying on expensive cloud computing resources.
The Ara240 DNPU includes integrated secure boot and a hardware root of trust (RoT) processor. This provides strong hardware-based security features for industrial, enterprise and embedded deployment environments.
The Ara240 DNPU supports PCIe Gen4 and USB 3.2 host interfaces. As a result, it can be easily integrated into edge computing platforms, PCs, notebooks and AI-enabled embedded systems.
This performance is presented as sufficient to run the latest AI workloads directly at the edge in industrial automation systems, autonomous robots, smart infrastructure, advanced HMI platforms and edge server design.
Mouser also highlighted the comprehensive capabilities of the Ara SDK for the Ara240 DNPU, saying it simplifies the development of a wide range of AI workloads for Ara240-based DNPUs and modules and accelerates AI application development.
As evidence, Mouser pointed to the fact that the Ara SDK includes a scalable AI compiler and supports a wide and continuously evolving range of model architectures, including CNNs (convolutional neural networks), FCNNs (fully connected neural networks), ViTs (vision transformers), LLMs and VLMs, as well as highly complex language models.
Mouser supplies the ARA2-M2-16G-GT AI accelerator module based on the Ara240 DNPU. The ARA2-M2-16G-GT adopts the M.2 M-Key design specification and features interface compatibility based on an x4 configuration and support for up to 4 PCIe Express 4.0 lanes.
The ARA2-M2-16G-GT is designed in the NGFF 2280 form factor and can be used on boards with M.2 M-Key PCIe functionality. As a result, it can be easily used as a co-processing device on the FRDM i.MX 95 and FRDM i.MX95 Pro development boards.
Source: TECHWORLD · Lee Gwang-jae
Original: https://www.epnc.co.kr/news/articleView.html?idxno=408444
References
This article was produced with the help of an automated content generation algorithm.
Source: TECHWORLD
View originalThis article was summarized and organized by BizCrush based on the original article from TECHWORLD. For exact quotations and full details, please refer to the original article.