Neural network processor based on systolic array

Abstract

The invention provides a neural network processor which comprises a control unit, a calculation unit, a data storage unit and a weight storage unit. The calculation unit controlled by the control unitacquires data and weights from the data storage unit and the weight storage unit to perform operation related to a neural network. The calculation unit comprises an array controller and a plurality of processing units connected in a systolic array mode, the data and the weights are input to a systolic array formed by the processing units in different directions, and the processing units simultaneously and parallelly process the data flowing through the processing units. The neural network processor can reach high processing speed, and the input data are repeatedly reused, so that high operation throughput rate can be realized under the condition of consuming small memory access bandwidth.

Description

technical field [0001] The present invention relates to neural network technology, in particular to neural network processor architecture. Background technique [0002] Deep learning has made major breakthroughs in recent years. The neural network model trained with deep learning algorithms has achieved remarkable results in image recognition, speech processing, intelligent robots and other application fields. The deep neural network simulates the neural connection structure of the human brain by building a model, and describes the data features hierarchically through multiple transformation stages when processing signals such as images, sounds, and texts. With the continuous improvement of the complexity of the neural network, the neural network technology has many problems in the actual application process, such as occupying a lot of resources, slow operation speed, and large energy consumption. The method of replacing traditional software computing with hardware accelera...

AI Technical Summary

Problems solved by technology

[0003] However, since the neural network processor is a computationally intensive and memory-intensive processor, on the one hand, the neural network model includes a large number of multiplication and addition operations and other nonlinear operations, which require the neural network processor to maintain high load operation to ensure the neural network The computing requirements of the network model; on the other hand, there are a large number of parameter iterations in the neural network computing process, and the computing unit needs to access a lot of memory, which greatly increases the bandwidth design requirements of the processor and increases memory access power consumption.

Images