Power efficiency and edge intelligence with new processor

Advantech has introduced the SOM-6820, a COM Express Compact Type 6 module powered by the Qualcomm Snapdragon X-Elite series processors. Featuring up to 12 cores and a 45W TDP, the SOM-6820 delivers up to 65% greater power efficiency at a fraction of the power than competitive platforms. Its integrated NPU supports up to 45 TOPS of AI performance. In addition to its powerful processing capabilities, the device supports up to 64GB of LPDDR5x memory and onboard UFS storage of up to 256GB, ensuring exceptional data processing performance. It also provides a wide range of I/O options, including 14 x PCIe, 4 x USB 3.0, and LAN connectivity, providing maximum configuration flexibility for diverse development requirements. Equipped with the company's QFCS (Quadro Flow Cooling System), the module allows easy assembly and quiet operation. The company further supports developers with comprehensive services, including UEFI BIOS, Windows 11, and the Edge AI SDK, enabling seamless migration between computing architectures. With its cutting-edge performance, ten-year product longevity, and full-service support, the SOM-6820 expands the possibilities for COM modules, advancing applications in medical imaging, machine vision, mission-critical systems, and humanoid robotics.

The SOM-6820, powered by the Qualcomm Snapdragon® X-Elite and X Plus series processors, delivers up to 12 cores, offering a 51% performance increase over similar x86 platforms, according to Qualcomm's report. It also combines exceptional AI capabilities via the Adreno GPU (up to 4.6 TFLOPS) and the integrated Hexagon NPU (up to 45 TOPS) to accelerate edge inference. The X-Elite's enhanced power efficiency and AI performance significantly improve inference quality and processing speed in applications such as medical imaging and machine vision. It also reduces the need for external graphics cards, boosting performance while lowering both cost and space requirements.

To complement the optimised SoC efficiency, the SOM-6820 features onboard LPDDR5x low-power memory, supporting up to 64GB (8448 MT/s), and onboard UFS 3.1 storage of up to 256GB for high-speed data access. With a wide operating temperature range of -40C to 85C, the device ensures stable performance in harsh environments. Designed for flexible configuration, it supports 14 x PCIe lanes, allowing connectivity for up to five end devices. Additional interfaces include 4 x USB 3.0 ports, 4 x SATA3 ports, and 1 x GbE LAN port.

To ensure performance stability, the SOM-6820 supports the company's Quadro Flow Cooling System (QFCS), allowing silent operation and sustained 100% performance under 45W and 60C conditions. Beyond hardware, it supports Windows 11, enabling seamless migration from x86 applications. It also features a UEFI BIOS with x86 usage scenarios in mind, simplifying hardware and firmware configuration through intuitive BIOS menus. For AI implementation, Advantech provides the Edge AI SDK, a rapid development toolkit that streamlines AI evaluation, development, and deployment. With these comprehensive services, adopting and migrating to the SOM-6820 becomes simpler and more efficient.