We’ve already seen M.2 cards based on one or more Intel Movidius Myriad X VPU with the likes of AAEON AI Core XM2280 M.2 card, but there’s now another option from Taiwan-based IEI Integration Corp with their Mustang-M2MB-MX2 card.
- AI Accelerators – 2x Intel Movidius Myriad X MA2485 VPU
- Dataplane Interface – M.2 BM Key
- Power Consumption – Around 7.5W
- Cooling – Active Heatsink
- Dimensions – 22 x 80 mm
- Temperature Range – -20°C~60°C
- Humidity – 5% ~ 90%
Just like other Myriad X devices, the card relies on Intel OpenVINO toolkit working on Ubuntu 16.04.3 LTS 64-bit, CentOS 7.4 64-bit or Windows 10 64-bit operating systems, and supporting AlexNet, GoogleNetV1/V2, MobileNet SSD, MobileNetV1/V2, MTCNN, Squeezenet1.0/1.1, Tiny Yolo V1 & V2, Yolo V2, ResNet-18/50/101 topologies, as well as TensorFlow, Caffe, MXNet, and ONNX AI frameworks.
The heatsink is really thick (~2 cm high), so it’s not something you’d just put in your laptop, and instead, it’s better suited to specialized systems such as the company’s FLEX-BX200-Q370 AI inference system based on Coffee Lake processor and equipped with M.2 and PCIe ports for such kind of cards.
Thanks to Rainer for the tip.
Jean-Luc started CNX Software in 2010 as a part-time endeavor, before quitting his job as a software engineering manager, and starting to write daily news, and reviews full time later in 2011.