Advantech Introduces Intel Arc A380E-Powered Industrial GPU for Enhanced AI Computing

Published by

teaser
Advantech, recognized for its contributions to intelligent IoT systems and embedded platforms, has recently unveiled the EAI-3101, an industrial-grade PCIe GPU card featuring the Intel Arc A380E. Designed for a five-year lifespan, this GPU card integrates 128 Intel Xe matrix AI engines, achieving an AI computational power of 5.018 TFLOPS. This performance effectively doubles that of the NVIDIA T1000 and incorporates advanced features such as ray tracing technology and Intel XeSS AI-upscaling, supporting resolutions up to 8K UHD and enhancing graphics performance by 50% compared to the NVIDIA T1000.

To facilitate rapid development in Vision AI, Advantech offers the Edge AI SDK, which significantly reduces development time by 40% when used in conjunction with Intel OpenVINO. The EAI-3101 is tailored for demanding applications in gaming, medical analysis, and video surveillance, featuring an optimized thermal design and an automatic smart fan. This product will be showcased at Embedded World 2024 in Nuremberg, Germany, from April 9th to 11th. The EAI-3101 leverages Intel Deep Link technologies, allowing for efficient workload optimization and power sharing between Intel Integrated GPUs and Arc GPUs, thereby maximizing AI computing capabilities. Compatible with Advantech systems and motherboards such as the DPX-S455, AIMB-278, and ARK-3534, this GPU card ensures optimal performance for a variety of applications.


87989776658


Equipped with four DP1.4a outputs, the EAI-3101 supports 8K UHD resolutions through Intel XeSS AI-upscaling technology, making it suitable for high-resolution applications in medical analysis and gaming casinos. It also offers comprehensive media codec and software API support, including AV1/VP9/DirectX, paving the way for advanced visual computing.

Advantech's Edge AI SDK and Intel OpenVINO toolkit streamline the AI development process, offering tools for inference benchmark evaluation, runtime SDK integration, model conversion, and technical consulting. This enables efficient AI inference application development and deployment across Windows 11/10 and Ubuntu 22.04 operating systems.


Share this content
Twitter Facebook Reddit WhatsApp Email Print