Giada LM2-100 AI Accelerator Module

The Giada LM2-100 AI Accelerator Module is a specialized edge AI inference card tailored for integration into embedded systems, industrial PCs, and single-board computers. Manufactured by Giada, this module, SKU LM2-100, belongs to the AI accelerator category and features the DeepX M1M Neural Processing Unit, providing up to 25 TOPS of INT8 inference capacity while optimizing for low power use.
 
Key technical features include the M.2 2242 B+M-Key form factor allowing compact deployment and a PCIe Gen3 x2 interface for high-speed connectivity to host platforms. It integrates 2GB of LPDDR4x dedicated memory to handle AI workloads locally without taxing system resources. The module supports prominent AI frameworks such as TensorFlow, TensorFlow Lite, ONNX, Keras, and PyTorch via the Dataflow compiler, facilitating versatile deployment of diverse neural network models.
 
Engineered for robust edge computing, the LM2-100 excels in applications like industrial quality control, intelligent surveillance systems, smart retail environments, drone vision, and robotics automation. Certified to operate reliably across a temperature range of -25°C to 65°C, it supports Windows and Linux OS platforms, making it a resilient and adaptable choice for advancing AI-powered industrial and IoT infrastructures.
 
DeepX M1M NPU with 25 TOPS AI performance
M.2 2242 B+M-Key form factor
PCIe Gen3 x2 host interface
2GB LPDDR4x memory for local AI workloads
Supports TensorFlow, ONNX, Keras and PyTorch
Operates from -25°C to 65°C