ADLINK Technology launched the first embedded MXM graphics module based on NVIDIA Ampere architecture, suitable for edge computing and AI
ADLINK Technology recently released the industry’s first embedded MXM graphics module based on the NVIDIA Ampere architecture, which is designed for edge-accelerated computing and AI workloads. The new embedded graphics module adopts a compact mobile PCI express (MXM) shape design, which can realize real-time ray tracing, AI accelerated graphics, and energy-efficient AI inference acceleration. These Modules can improve responsiveness, accuracy and reliability for mission-critical and time-sensitive applications in the medical, manufacturing, and transportation fields.
● ADLINK’s embedded MXM graphics module has a compact and energy-efficient MXM shape design, which can achieve high-performance GPU acceleration, allowing many vertical markets such as medical, manufacturing, and transportation to adopt edge computing and embedded AI.
● Rugged design can withstand extreme temperature, shock and vibration, and corrosion under severe conditions.
Cai Yuli, Product Director of Embedded Platforms and Modules Division of ADLINK’s Platform Product Center, said: “Compute and AI workloads are shifting from the cloud to edge devices. In addition to shortening response time and increasing security, it also reduces communication costs. Edge needs The data being processed is increasing, and the performance requirements of edge devices are gradually increasing, but the power budget has not been adjusted accordingly. The NVIDIA Ampere architecture provides a major breakthrough in performance and energy efficiency, and promotes edge devices in general computing, image processing and reconstruction , As well as AI reasoning ability are all improved.”
Scott Fitzpatrick, vice president of product marketing at NVIDIA, said: “The NVIDIA Ampere architecture provides breakthrough performance and features, integrating the latest generation of RT cores, Tensor cores, CUDA cores, PCIe Gen 4 and NVIDIA video codecs. With rendering and Simulating a wide range of applications in all walks of life, NVIDIA’s latest embedded solutions can provide up to 2 times the rendering performance, 2 times the FP32 throughput, and hardware-accelerated video codecs to cope with the substantial increase in graphics and computing workloads. “
ADLINK’s embedded MXM GPU module is based on the NVIDIA Ampere architecture and is equipped with up to 5120 CUDA cores, 160 Tensor cores and 40 RT cores, supports PCIe Gen 4 and up to 16GB GDDR6 memory, and has a TGP of up to 115 watts. These modules can satisfy applications that are computationally intensive, require high graphics, and require high memory. Its size is one-fifth the size of a full-height, full-length PCI Express graphics card, and it is more robust, able to withstand extreme temperatures, shocks and vibrations, and is corrosion-resistant, suitable for edge environments where size, weight, and power are limited. These embedded graphics modules provide five-year life support. Developers, solution architects, and system integrators can trust the stable supply during the product life cycle, and innovate around new solutions with peace of mind.
Applications of the embedded MXM graphics module include:
● Medical and health: speed up the image reconstruction of mobile X-ray, ultrasound and endoscopic systems
● Transportation: real-time object detection on railway or airport runways to improve transportation safety
● Retail and logistics: navigation and route planning for autonomous drones and mobile robots (AMR) to assist in the last step, last-mile delivery
● Aerospace and defense: time-sensitive and mission-critical command, control, communications, computer (C4) intelligence, surveillance and reconnaissance (ISR) applications