ניו-טק מגזין | יוני 2026 | מהדורה דיגיטלית

DX-M1 M.2 LPDDR5x2 for Every AIoT The DEEPX DX-M1 M.2 module brings server-grade AI inference directly to edge devices. Delivering 25 TOPS of performance at just 2-5W, the module achieves 20x better performance efficiency (FPS/W) than GPGPUs while maintaining GPU-level AI accuracy.

DX-M1 M.2 LPDDR5x2 for Every AIoT The DEEPX DX-M1 M.2 module brings server-grade AI inference directly to edge devices. Delivering 25 TOPS of performance at just 2-5W, the module achieves 20x better performance efficiency (FPS/W) than GPGPUs while maintaining GPU-level AI accuracy.

“Integrating powerful AI vision processing and essential core features into a single chip DEEPX drives innovative edge AI solutions for diverse intelligent systems.”

“Integrating powerful AI vision processing and essential core features into a single chip, DEEPX drives innovative edge AI solutions for diverse intelligent systems.”

DX-M1 M.2 LPDD5Rx2: Functional Block Diagram

DX-M1 M.2 LPDD5Rx2: Functional Block Diagram

QSPI Flash (Nand) 1Gbit

UART SWD

Key Features Type: AI Accelerator

QSPI Flash (Nand) 1Gbit

UART SWD

50MHz

DQ 16bit x 2

XTAL

LPDDR5 #0.1 1GB + 1GB Key Features Type: AI Accelerator Form Factor: M.2 M Key (22 x 80 mm) Interface: PCle Gen.3 x4 Memory: 4GB LPDDR5, QSPI 1Gbit NAND Flash Host HW: x86, ARM Based Architecture LPDDR5 #2, 3 1GB + 1GB Support DXNN® SDK DXNN® SDK is a comprehensive SW development environment for deploying AI on DEEPX NPUs. It integrates tools for compiling, optimizing, simulating, and inferring the latest AI models, such as YOLO, ViT, and VLMs. And it provides an optimized, ready-to-use environment as the DX-All Suite package to support fast and efficient AI development.

50MHz

DQ 16bit x 2

XTAL

Target Applicatio • 3D Sensing & Stereo Came • AI NVR (Network Video Rec Support DXNN® S DXNN® SDK is a comprehens environment for deploying AI It integrates tools for compilin simulating, and inferring the l such as YOLO, ViT, and VLMs an optimized, ready-to-use e DX-All Suite package to supp AI development. Form Factor: M.2 M Key (2 Interface: PCle Gen.3 x4 Memory: 4GB LPDDR5, QS Host HW: x86, ARM Based • Video Conferencing Camer • CMS (Camera Monitoring S • Autonomous Robotic Platfo • Drones • AR/VR • SBC (Single Board Compute • ADAS/AD • AI CCTV • Robotics • Automotive

M.2 PCIe Gen3 x4

LPDDR5 #0.1 1GB + 1GB

DQ 16bit x 2

M.2 PCIe Gen3 x4

3 Color LED

DQ 16bit x 2

LPDDR5 #2, 3 1GB + 1GB

3 Color LED

Specifications Features

Specifications

AI Accelerator INT8 Performance

Details 25 TOPS

Specifications

Processor

Feature

AI Accelerator

Details

Signal Interface AI Accelerator

PCI Express

PCle Gen.3 x4 / Bandwidth: 4GB/s *Compatible to PCIE x1

Processor

INT8 Performance 25 TOPS

Feature

Details

Power

Power Consumption

2W min., 5W max. for DEEPX supported models

PCle Gen.3 x4 / Bandwidth: 4GB/s *Compatible to PCIE x1 -25 ~ 85°C (Throttling) -25 ~ 65°C (Non_Throttling)

Signal Interface PCI Express

Processor

INT8 Performance 25 TOPS

Operating

Temperature

Power

Power Consumption 2W min., 5W max. for DEEPX supported models

Humidity PCle Gen.3 x4 / Bandwidth: 4GB/s *Compatible to PCIE x1

Environment Signal Interface PCI Express

40 °C @ 85% relative humidity (non-condensing)

-25 ~ 85°C (Throttling) -25 ~ 65°C (Non_Throttling) Heatsink (Option)

Operating

Temperature

Thermal Solution Power Consumption 2W min., 5W max. for DEEPX supported models Cooling

Power

Environment

Humidity -25 ~ 85°C (Throttling) -25 ~ 65°C (Non_Throttling) Form Factor

40 °C @ 85% relative humidity (non-condensing) M.2 2280 (Key M)

Target Applications • 3D Sensing & Stereo Cameras • AI NVR (Network Video Recorder)

Operating

Temperature

Physical

Thermal Solution Cooling Dimensions

22mm x 80mm x 4.1mm

Heatsink (Option)

Environment

Humidity

40 °C @ 85% relative humidity (non-condensing)

Power Range

M.2 2280 (Key M) 3.3V ± 5%

Form Factor

Thermal Solution Cooling

Windows Heatsink (Option) Linux M.2 2280 (Key M)

Windows 11, 10 64 bit Ubuntu 22.04, 20.04 LTS Support Yocto Project and Docker Support TensorFlow, TensorFlow Lite, ONNX, Keras, PyTorch by Dataflow complier converted • AI CCTV • Robotics • Automotive • Video Conferencing Cameras

Physical

Dimensions

22mm x 80mm x 4.1mm

Software Support Form Factor

Power Range

3.3V ± 5%

Framework 22mm x 80mm x 4.1mm

Physical

Dimensions

Windows

Windows 11, 10 64 bit

• CMS (Camera Monitoring System) • Autonomous Robotic Platforms • Drones • AR/VR • SBC (Single Board Computer) • ADAS/AD

System Support Power Range

CPU Platform

x86, ARM Based Architecture

3.3V ± 5%

Software Support

Ubuntu 22.04, 20.04 LTS Support Yocto Project and Docker

Linux

Framework Amir Sherman Sales & Business Development EMEA +972 52 2240811 +49 173 3232288 as@deepx.ai Windows 11, 10 64 bit Ubuntu 22.04, 20.04 LTS Support Yocto Project and Docker System Support CPU Platform

Windows

Support TensorFlow, TensorFlow Lite, ONNX, Keras, PyTorch by Dataflow complier converted

Software Support

Linux

x86, ARM Based Architecture

Support TensorFlow, TensorFlow Lite, ONNX, Keras, PyTorch by Dataflow complier converted

Framework

System Support CPU Platform

x86, ARM Based Architecture

DEEPX HQ 5F, 20 Pangyoyeok-ro 241beon-gil, Seongnam-si, Gyeonggi-do, South Korea

USA 1735 Technology Drive Suite 740.San Jose, CA U.S.A

Taiwan No. 66-10, Yucheng St., Nangang Dist., Taipei City 115012, Taiwan (R.O.C.)

Europe Tel-Aviv Israel

Buy Now! - Contact Sal