Neural network acceleration hardware architecture and method for quantization bit width dynamic selection

Abstract

The invention discloses a neural network acceleration hardware architecture and method for quantization bit width dynamic selection. The hardware architecture comprises a global storage module, a data scheduling module, a local storage module, a dynamic quantization prediction controller and a calculation unit array. The method comprises the following steps: dividing neurons of a feature map in a network by taking a block as a unit, continuing to divide in a spatial dimension by taking a group as a unit in each block, and defining the group as a minimum unit for executing a dynamic quantization operation; configuring a trainable threshold parameter for each block in all feature maps of the target network, and determining upper and lower bounds of selectable sparseness of each block according to a given basic quantization bit width and a target bit width total amount constraint; establishing a dynamically quantized neural network training and reasoning model, dividing reasoning into two parts of high-precision calculation and low-precision calculation, and judging whether to execute the low-precision calculation or not according to a result of the high-precision calculation; on the premise of ensuring that the precision is not lost, the reasoning time in actual hardware is reduced as much as possible.

Description

technical field [0001] The invention relates to the technical fields of embedded data intelligent processing and artificial intelligence; in particular, it relates to a neural network acceleration hardware architecture and method for dynamic selection of quantization bit width. Background technique [0002] In recent years, deep learning has been widely used in data processing fields such as images and voices. Its powerful data analysis, prediction, modeling and classification capabilities also have unique advantages in dealing with complex problems in the field of digital grids, including through images, voices, etc. Recognition technology realizes image analysis of transmission line inspection, intelligent monitoring and alarm of power distribution room and cable pipe gallery, etc., and can achieve more accurate assessment of safety risks around power grid equipment. In addition, intelligent analysis of data in deep learning can be applied to power grids Control text knowl...

AI Technical Summary

Problems solved by technology

But it still does not take into account the differences between different channels and neurons in the same layer

Images