Embedded Computing Design

DFI’s ECX700-AL is a Ruggedized Embedded System, With IP67/69K waterproof and dust-proof design to withstand harsh environments and any type of climate. Stable performance is ensured with its wide operating temperature of -40 to 70 degrees Celsius. It is also equipped with a smart vent that drains water automatically. High quality wireless transmission is ensured with a high gain antenna to maximize signal strength and an external SIM slot for easy accessibility. Versatile I/O, including selective display ports (HDMI/VGA) and a combo port that supports 2 CANs and 2 COMs, allows for integrators to meet demands of various applications. 

The OnLogic Factor 202 leverages the Raspberry Pi Compute Module 4 (CM4) to deliver a compact, ARM-powered, purpose-built industrial controller. The system provides the requisite intelligence and connectivity for applications including Smart Buildings, Advanced Manufacturing, IIoT Edge, and SCADA deployments. The system includes an onboard touchscreen to facilitate Human Machine Interface (HMI), and a choice of either Raspberry Pi or Ubuntu OS. Customers may also select a variant of the system, called the IGN802, that comes preloaded with Inductive Automation's Ignition Edge, a leading SCADA application. 

JACKSON works with the NVIDIA® Jetson™ Orin™ Nano and Orin NX GPU modules to deliver high performance, I/O-rich solutions for AI-at-the-edge applications in a compact size. Jackson’s Orin-optimized design provides access to all module features to deliver up to double the I/O capability of competing carrier boards. Core I/O and expansion includes dual Ethernet ports, dual 4-lane CSI camera connectors, 4 USB 3 / USB 2 ports, and dual M.2 sockets (1 2242/2280 M key and 1 2230 E key). Jackson also has 2 RS-232/485 ports, 16 3.3V/5V GPIO lines, CAN 2.0, and HDMI 2.0 a/b with audio support. An expansion connector provides PCIe, USB3, and power to connect I/O daughterboards. Customers can easily enhance Jackson with standard or custom daughterboards without paying high NRE charges for a full-custom solution.

BrickML is a standalone device for measuring and monitoring machine health with machine learning (ML). It tracks current (MCSA), vibration, temperature, and audio via its integrated sensors and onboard computer. Using advanced ML algorithms, BrickML analyzes data on the device itself to detect subtle changes in machine behavior, predicting potential failures before they occur. The compact design allows for easy integration into existing systems, providing real-time data analysis and actionable insights, right at the edge. It is fully integrated into the Edge Impulse platform, allowing users to quickly and easily create AI algorithms for their specific use cases. 

LIPSedge F110 is an Industrial-ready AGX-Xavier based PCIe based edge accelerator for GigE/ PoE RGB-D camera with GenICam. It features 2 robust PoE+ ports with PSE (up to 30 watts) and can run at PCIe endpoint mode. It supports the high bandwidth data transfer with PCIe 4.0 up to 16GB/s and low latency direct data transmission to GPU via RDMA without taking up CPU. LIPSedge F110 features Cleverboot Interconnect for multi-OS systems without use of a hypervisor and allows programmable edge computing with NVIDIA CUDA and OpenCL. It is compatible with LIPSedge 3D Camera and SDK.

ASUS IoT PE6000G is the industry’s first edge AI system that supports up to 450W GPU for high throughput, demanding GPU computing applications. Powered by 12th Gen Intel Core™ processor and paired with the Intel R680E chipset, PE6000G supports the latest PCI Express® (PCIe®) 5.0 standard to accommodate graphics cards with up to 450 W TDP, making it perfect to meet the requirements for enabling real-time AI inference at the edge. 

Aetina's AI-MXM-H84A MXM module features four Hailo-8™ AI processors, providing up to 104 TOPS of AI performance with power consumption of 25W. The 3.1 Type B MXM module supports TensorFlow and ONNX, able to function well with the operating temperature from 0°C to +70°C. 

Edge AI Vision applications become more and more common helping various industries to resolve real-life issues, whether is it detecting manufacturing defects or providing intelligent traffic analysis. Prototyping, however, can be very troublesome and time-consuming, as the developer has to spend time selecting vision system components including HW module, camera sensor, cabling, and more. And then have to integrate and debug all needed drivers and packages as well as OS or run AI vision applications. With this unique, all-in-one dev kit, prototyping and testing become easy and straightforward, just open the box and start building apps instantly, all you need in terms of HW and SW is provided and pre-installed. 

The Lattice Avant platform will be the foundation for multiple new feature-rich mid-range FPGA families and is purpose-built to expand Lattice’s power efficient architecture, small size, and performance leadership more parts of the FPGA market. Designed for scalability, Lattice Avant provides the base technology for a multitude of FPGA variants optimized for different applications across the Industrial, Automotive, Communications, and Computing markets. The distributed resources on our FPGAs’ fabric can process data in parallel at a lower operating frequency that other types of processors and, when compared with similar class competitive devices, Avant-based FPGAs are up to 2.5X more power efficient.

Camera modules are critical for AI vision applications, providing essential information to the algorithm. The EV2U-GOM1 camera module's advantages, such as its compact size, power efficiency, and high bandwidth, make it ideal for AI hosts like the NVIDIA Jetson and AMD Xilinx KV260. Its superior fisheye calibration ISP expands fisheye lens images into several modes, helping AI hosts focus on inference tasks. Without a camera module, the AI algorithm would be blind and unable to perform tasks effectively. The camera module's high-quality images are quickly analyzed by the AI algorithm, extracting meaningful information for a more efficient system.

The DynaCOR 40-36 is a liquid-cooled, rugged and compact edge AI server certified for automotive applications, enabling edge inference and training for ADAS and autonomous driving development. It can be configured with up to 4x NVIDIA GPUs to offer unprecedented performance for a fanless system: up to 237 TFLOPS for Deep Learning precision, up to 29.6 TFLOPS for single precision and up to 14.8 TFLOPS for double precision. Combined with Eurotech portfolio of high performance data loggrs and networking systems, it enables data cententer-class architectures to perform in-vehicle training and inference of AI algorithms for autonomous driving. 

AI EdgeLabs is a powerful and autonomous cybersecurity AI solution that helps organizations respond immediately to attacks and protect their Edge/IoT critical infrastructure from malware, ransomware, DDoS, and other threats. This next-generation XDR platform equips security teams with multi-layered detection and prevention software, advanced network visibility, and asset discovery combined with advanced firewalling and SOAR elements. AI EdgeLabs provides confidence, scalability, and a robust security posture for clients thanks to the platform’s comprehensive capabilities which include easy integration, automated incident remediation, in-depth visibility, data security, modular architecture, and a proprietary EdgeLabs sensor for network telemetry and threat prevention software.

ADLINK integrates Intel’s latest line of powerful, discrete graphics - Intel® ARC™ A-Series Graphics Processor Mobile (Alchemist) — on the embedded MXM form factor. It leverages Intel’s well-established graphics ecosystems, such as AI OpenVINO, Media SDK management tools, that edge developers have enjoyed and relied on for years. This isn’t a mere new line of GPUs, but one that makes your migration from integrated to discrete graphics seamless and fully transparent. 

Supermicro's X13 Hyper-E brings top performance and flexibility of Supermicro’s flagship Hyper series to the edge with a dual 4th Gen Intel® Xeon® Scalable Processors (up to 60 cores each) in a short -depth of 574mm (22.6”) 2U design for edge data and telco deployments. The system allows up to 8TB ECC DDR5 DRAM in 32 sockets. 8 PCIe Gen 5 expansion sockets (form 8x8 to 4x16) with three FHHL slots. Connectivity with 10, 25 and 100 Gbe network adapters. 6x 2.5” hot swap NVME drive bay in Raid 0/1/5/10 configuration. Redundant front access 2000W AC or DC power supplies. 

NET-III 2I640DW is a Palm size DIN-Rail computer with dimensions measuring just 35 x 110 x 135 mm. Features with Intel Elkhart Lake Atom® processor and M.2 Hailo-8 AI accelerator module make it as the excellent computing system to fulfill the needs of Edge AI applications. In addition, NET-III 2I640DW is packed full of interfaces including dual independent HDMI display, 2 port USB 3.0, 3 port 2.5G LAN and 2 COM port. NET-III 2I640DW a trusted solution for projects that require space-constrained edge computing applications for intelligent robotic control, smart traffic, smart medical, visual inspection and analysis.

ADATA’s industrial temperature-rated ISDD33K SD card is the first product that integrates high capacities up to 512GB and wide temperature operability from -40 ºC to 85 ºC in the market. It employs 112-layer 3D TLC Flash and sports a 3K P/E cycle rating for great durability.

DDR5, the latest DRAM generation, offers significant improvements in all areas. With 4x increased capacity, 2x speed, and lower voltage than the previous generation, DDR5 is a clear upgrade. Innodisk has gone a step further by designing a DDR5 Ultra Temperature SODIMM module that can operate in extreme temperatures ranging from -40°C to 105°C.

Innodisk's nanoSSD PCIe 4TE3 is the first PCIe Gen.4 x4 NVMe BGA SSD. It is designed in the standard M.2 Type 1620 BGA form factor and features a PCIe interface with up to 8GB/s bandwidth. With a built-in 12nm process NAND Flash controller and industrial-grade 3D TLC NAND Flash, it provides high performance, capacity, endurance, and reliability while consuming low power.

Apacer’s patented SV25T Transformed SSD series connects a M.2 2242 SSD to a rugged connector and a value-adding M.2 2238 PCB, while still remaining within the standard dimensions of a M.2. 2280 SSD. One value-added function module is the backup and recovery technology, the Transformed SSD OOB (Out-of-band) Module offers users a separate channel used for transmitting data outside the main data channel for monitoring, troubleshooting, or emergency recovery.

Reliance Sense is a flash file system for storing structured data in NOR flash. It efficiently makes use of the byte-addressable mode of NOR flash to minimize flash wear. The file system has a circular design to suit continuous recording workloads, while configuration data is stored according to a linear design.