公开(公告)号：US20220236909A1

公开(公告)日：2022-07-28

申请号：US17722890

申请日：2022-04-18

Applicant: Huawei Technologies Co., Ltd.

Inventor： Xingcheng Hua ,  Zhong Zeng ,  Leibin Ni

IPC: G06F3/06 ,  G06N3/063

Abstract: A neural network computing chip and a neural network computing method are provided. The computing chip includes a translation circuit and a computing circuit. After input data is processed by the translation circuit, a value of each element of the data input into the computing circuit is not a negative number, thereby meeting a value limitation condition imposed by the computing circuit on the input data. Therefore, a calculation result can be obtained by performing calculation for only one time, and there is no need to perform calculation for two times. Therefore, neural network computing efficiency is improved, and a delay of a neural network operation is reduced.

公开(公告)号：US11853594B2

公开(公告)日：2023-12-26

申请号：US17722890

申请日：2022-04-18

Applicant: Huawei Technologies Co., Ltd.

Inventor： Xingcheng Hua ,  Zhong Zeng ,  Leibin Ni

IPC: G06F3/06 ,  G06N3/063

CPC classification number: G06F3/0655 ,  G06F3/0604 ,  G06F3/0679 ,  G06N3/063

Abstract: A neural network computing chip includes a translation circuit and a computing circuit. After input data is processed by the translation circuit, a value of each element of the data input into the computing circuit is not a negative number, thereby meeting a value limitation condition imposed by the computing circuit on the input data.