Memory chip capable of executing artificial intelligence operation, and operation method thereof

Abstract

The invention provides a memory chip capable of executing artificial intelligence operation, and an operation method thereof. The memory chip includes a memory array and an artificial intelligence engine. The memory array is used for storing input characteristic value data and a plurality of weight data. The input feature value data include a plurality of first subsets, and each weight data includes a plurality of second subsets. The artificial intelligence engine includes a plurality of feature value detectors and is configured to access the memory array to obtain input feature value data andweight data. Each feature value detector selects at least one second subset from the corresponding weight data as a selected subset according to the weight index, and the feature value detector executes neural network operation according to the selected subset and the corresponding first subset.

Description

technical field [0001] The present invention relates to a memory architecture, in particular to a memory chip capable of performing artificial intelligence (AI) operations and an operation method thereof. Background technique [0002] With the evolution of artificial intelligence (AI) computing, the application range of artificial intelligence computing is becoming more and more extensive. In addition to being used in cloud servers equipped with high-speed graphics processing units (Graphics Processing Unit, GPU), it can also be used in For example, small portable electronic devices such as mobile phones, tablet computers, and Internet of Things (IoT) devices. In this way, convolutional neural network operations such as image analysis can be performed on the device through the neural network model to improve the performance of the operation. [0003] For portable electronic devices, reducing energy consumption is a well-known issue. However, performing convolutional neural...

AI Technical Summary

Problems solved by technology

However, performing convolutional neural network operations on portable electronic devices can incur considerable energy consumption due to the large number of parameters required for many convolutional neural networks

In small portable electronic devices, the cache memory on the processing chip is not enough to store so many parameters, so these parameters must be read from the external DRAM (Dynamic Random Access Memory) outside the artificial intelligence computing engine every time a calculation is performed. Access Memory, DRAM) is moved to the processing chip once, resulting in high energy consumption

Images