Deep neural network accelerator fault handling method and device

Abstract

The embodiment of the invention relates to a deep neural network accelerator fault handling method and device. The method comprises the following steps: generating a plurality of groups of input datafor testing the deep neural network and an output result corresponding to the input data; opening an accelerator data path of the depth neural network to a test channel, and sending the input data tothe accelerator through the test channel to complete the test operation; reading operation data of a plurality of operation units which are operated and stored in corresponding registers; matching theoperation data with a corresponding output result; If there is no match, the operation unit corresponding to the operation data is determined as a fault unit, and the fault unit is turned off, so that the DNN accelerator has the self-test ability, and the manufacturing fault is quickly tested and eliminated in the production stage, the test cost is reduced, and the available chip output is improved.

Description

technical field [0001] Embodiments of the present invention relate to the technical field of deep neural networks, and in particular, to a method and device for processing faults of deep neural network accelerators. Background technique [0002] With the breakthrough development of Deep Neural Network (DNN) in speech recognition and image recognition tasks, more and more hardware accelerator cores (Neural-network Processing Unit, NPU) dedicated to accelerating deep neural networks have been born. ). At the same time, the calculation amount of DNN also increases sharply with the application mode, calculation model and data processing volume, which makes the design scale of deep neural network accelerators become larger and larger, with more and more computing cores, regardless of hardware size or important To a certain extent, it has become the computing core with the same status as CPU and GPU, but its corresponding testing methods have not been developed accordingly, and t...

AI Technical Summary

Problems solved by technology

2. The consistent template leads to the encapsulation of operations. A single step (single instruction operation) is a set of operations. Once a hardware failure occurs, it is impossible to determine which part of the original error occurs.

[0005] Therefore, there is a lack of testing schemes applied to DNNs in the prior art

Images