TLM microstructure for GPU hardware image processing convolution filtering system

Abstract

The invention relates to the technical field of computer hardware modeling, in particular to a TLM microstructure for a GPU hardware image processing convolution filtering system. The TLM microstructure oriented to the GPU hardware image processing convolution filtering algorithm comprises convolution kernel loading, convolution parameter initialization, convolution data storage, convolution filtering calculation and pixel collection after convolution. According to the method, the function and the implementation structure of the image processing convolution filtering algorithm based on the TLMmodel are realized, the problem of storage algorithm function verification for replacing the TLM microstructure with GPU hardware sub-textures is solved, and RTL design and development are effectively accelerated.

Description

technical field [0001] The invention relates to the technical field of computer hardware modeling, in particular to a TLM microstructure oriented to a GPU hardware image processing convolution filtering system. Background technique [0002] In the design and development of a graphics processor chip (hereinafter referred to as GPU), the correctness and efficiency of algorithms are important factors that determine the function and performance of the GPU. Convolution filtering is an important function in image processing. GPU hardware implementation of convolution function needs to comprehensively consider the convolution mode supported by OpenGL API, convolution kernel, convolved data, and DDR storage space limited by hardware architecture, original data storage and Problems such as the coordination of data addresses during convolution operations, and the unification of outputs in different convolution modes. However, when the GPU chip uses RTL to implement the above algorith...

AI Technical Summary

Problems solved by technology

However, when the GPU chip uses RTL to implement the above algorithm details, the hardware logic scale is huge, and it is difficult to verify and debug at the RTL stage.

Images