1 / 2

Kneron NPU IP

Kneron NPU IP Series are neural network processors that have been designed for edge devices. These processors provide high computing performance with low power consumption and are small in size. Kneron NPU IP Series can be applied to smart homes, smart surveillance, smartphones, and wearable devices that have high requirement for low power and space. The entire product consumes under 0.5W and can even drop below 5mW for specific applications.
High energy efficiency | Kneron NPU IP | Kneron - Leading the Way in Edge AI

High energy efficiency

All series reach higher than 1.5 TOPS/W.
Support mainstream deep learning frameworks | Kneron NPU IP | Kneron - Leading the Way in Edge AI

Support mainstream deep learning frameworks

Caffe, Keras, TensorFlow, and ONNX etc.
Low power consumption | Kneron NPU IP | Kneron - Leading the Way in Edge AI

Low power consumption

Under 0.5W and can reach as low as 5 mW for specific applications.
An integrated AI solution | Kneron NPU IP | Kneron - Leading the Way in Edge AI

An integrated AI solution

Include hardware IP, compiler, and model compression solutions.
Deep compression technology | Kneron NPU IP | Kneron - Leading the Way in Edge AI

Deep compression technology

Compresses not only models but also data and coefficients during computing to reduce memory use. 
Filter decomposition and convolution acceleration | Kneron NPU IP | Kneron - Leading the Way in Edge AI

Filter decomposition and convolution acceleration

Divides a large-scale convolutional block into a number of smaller ones for parallel computing and then integrates and accelerates the results.
CNN model support optimization | Kneron NPU IP | Kneron - Leading the Way in Edge AI

CNN model support optimization

Supports diverse CNN models, including Vgg16ResnetGoogleNetYOLOTiny YOLOLenetMobileNet, and Densenet with model specific performance optimization.
Interleaving computation architecture | Kneron NPU IP | Kneron - Leading the Way in Edge AI

Interleaving computation architecture

Enables parallel convolution computing and pooling to improve overall performance. The convolution layer can support both 8- and 16-bits fixed points concurrently.
Adaptive data structure | Kneron NPU IP | Kneron - Leading the Way in Edge AI

Adaptive data structure

Adjusts the data structure dynamically to improve MAC efficiency depending on changing demand.
Dynamic storage resource configuration | Kneron NPU IP | Kneron - Leading the Way in Edge AI

Dynamic storage resource configuration

Allows more efficient resource allocation between shared memory and operational memory. Increases storage resource utilization without affecting performance.