Chinese Academy of Sciences

Huawei Technologies Co., Limited

Yinhe Han

Ying Wang

Haobo Xu

Xiaowei Li

Binzhang Fu

Hang Lu

Hang Xiao

Shiqi Lian

Feng Min

The present invention provides a processing system for a binary weight convolutional neural network. The system comprises: at least one storage unit for storing data and instructions; at least one control unit for acquiring the instructions stored in the storage unit and sending out a control signal; and, at least one calculation unit for acquiring, from the storage unit, node values of a layer in a convolutional neural network and corresponding binary weight value data and obtaining node values of a next layer by performing addition and subtraction operations. With the system of the present invention, the data bit width during the calculation process of a convolutional neural network is reduced, the convolutional operation speed is improved, and the storage capacity and operational energy consumption are reduced.

Processing system and method for binary weight convolutional neural network

Disclosed are a weight data storage method and a convolution computation method that may be implemented in a neural network. The weight data storage method comprises searching for effective weights in a weight convolution kernel matrix and acquiring an index of effective weights. The effective weights are non-zero weights, and the index of effective weights is used to mark the position of the effective weights in the weight convolution kernel matrix. The weight data storage method further comprises storing the effective weights and the index of effective weights. According to the weight data storage method and the convolution computation method of the present disclosure, storage space can be saved, and computation efficiency can be improved.

Weight data storage method and neural network processor based on the method

The present invention relates to a weight management method and system for neural network processing. The method includes two stages, i.e., off-chip encryption stage and on-chip decryption stage: encrypting trained neural network weight data in advance, inputting the encrypted weight into a neural network processor chip, and decrypting the weight in real time by a decryption unit inside the neural network processor chip to perform related neural network calculation. The method and system realizes the protection of weight data without affecting the normal operation of a neural network processor.

Weight management method and system for neural network processing, and neural network processor

An inverse kinematics solution system for use with a robot, which is used for obtaining a joint angle value corresponding to a target pose value on the basis of an inputted target pose value and degree of freedom of a robot and which comprises: a parameters initialization module, an inverse kinematics scheduler, a Jacobian calculating unit, a pose updating unit and a parameters selector. The system is implemented by means of hardware and may quickly obtain motion parameters, which are used for controlling a robot, while reducing power consumption.

Inverse kinematics solution system for use with robots

A task allocation method, a chip are disclosed. The method includes: determining the number of threads included in a to-be-processed task; determining, in a network-on-chip formed by a multi-core processor, a continuous area formed by routers-on-chip corresponding to multiple continuous idle processor cores whose number is equal to the number of the threads; when the area is a non-rectangular area, determining an extended area extended from the non-rectangular area; and when predicted traffic of each router-on-chip that is connected to a processor core in the extended area does not exceed a preset threshold, allocating the multiple threads of the to-be-processed task to the idle processor cores in the non-rectangular area. According to the task allocation method provided in the embodiments of the present invention, problems of large hardware overheads, a low network throughput, low system utilization are avoided.

Method, apparatus, and network-on-chip for task allocation based on predicted traffic in an extended area

A task allocation method, a chip are disclosed. The method includes: determining a number of threads included in a to-be-processed task; determining, in a network-on-chip formed by a multi-core processor, a continuous area formed by routers-on-chip corresponding to multiple continuous idle processor cores whose number is equal to the number of the threads; if the area is a non-rectangular area, determining a rectangular area extended from the area; and if predicted traffic of each router-on-chip that is connected to a non-idle processor core and in the extended rectangular area does not exceed a preset threshold, allocating the multiple threads of the to-be-processed task to the idle processor cores in the area. According to the task allocation method provided in the embodiments of the present invention, problems of large hardware overheads, a low network throughput, low system utilization are avoided.

Growing community of inventors

Beijing, China

Yinhe Han