AlphaICs, a Milpitas CA-based AI fabless semiconductor company that develops advanced inference and learning technologies, announced the availability of engineering samples of “Gluon” – a coprocessor of 8 TOPS Edge AI inference for surveillance, industrial, retail and automotive customers. , and industrial IoT verticals that deliver best-in-class FPS/Watt performance.
Gluon will come with a full SDK (Software Development Kit) that allows easy deployment of neural networks. The advanced state-of-the-art inference chip offers customers the ability to add AI capability into current X86/ARM-based systems, resulting in significant cost savings.
Gluon provides industry-leading fps/watt performance for neural network classification and detection – 32 frames per second (FPS)/watt for Yolo-V2, a leading object detection model, and 22 frames per second ( FPS)/Watt for VGG-19, an advanced classification model.
Produced using the 16nm FinFET process at TSMC
Gluon is currently being sampled for early customers to enable the development of their vision applications. It is designed for OEMs and solution providers targeting vision market segments such as surveillance, industrial, retail, industrial IoT, and edge gateway manufacturers.
Gluon, an 8 TOPS Edge AI coprocessor, is produced using TSMC’s 16nm FinFET process. This chip accelerates deep learning neural network models for classification, detection and segmentation and focuses on vision applications. Gluon integrates PCIe and LPDDR4 interfaces to enable high-speed transfers to host processors and DRAM, respectively.
Backed by many innovations, Gluon provides first-class references:
*153 frames per second with Yolo-V2, a leading object detection model (416x416x3 frame size) at 4.73 watts.
*79 fps with VGG-19, a leading rating model (224x224x3) at 3.6 watts.