Matrix convolution calculation method, interface, coprocessor and system based on RISC-V architecture

Abstract

The invention discloses a set based on RISC-. According to the method and system complete mechanism of the instruction, the interface and the coprocessor for matrix convolution calculation of the V instruction set architecture, traditional matrix convolution calculation is efficiently achieved in a software and hardware combined mode, and RISC-is utilized. Extensibility of V instruction sets, a small number of instructions and a special convolution calculation unit (namely a coprocessor) are designed; the memory access times and the execution period of a matrix convolution calculation instruction are reduced, the complexity of application layer software calculation is reduced, the efficiency of large matrix convolution calculation is improved, the calculation speed of matrix convolution isincreased, flexible calling of upper-layer developers is facilitated, and the coding design is simplified. Meanwhile, RISC-is utilized. The processor designed by the V instruction set also has greatadvantages in power consumption, size and flexibility compared with ARM, X86 and other architectures, can adapt to different application scenes, and has a wide prospect in the field of artificial intelligence.

Description

technical field [0001] The invention relates to matrix convolution calculation, in particular to a matrix convolution calculation method, interface, coprocessor and system based on RISC-V architecture. Background technique [0002] With the development of artificial intelligence, image recognition, neural network, deep learning and other technologies have been widely and deeply applied. There are a large number of matrix convolution calculations in the underlying implementation. Matrix convolution is the most basic and core algorithm of these technologies. The calculation of matrix convolution is the multiplication and accumulation of a large amount of data. The traditional C / C++ language implementation requires multi-layer loop calculations, multiple memory accesses to read data, and runs on systems with traditional architectures. The calculation speed, power consumption and volume It has been unable to meet the requirements of modern artificial intelligence. [0003] The ...

AI Technical Summary

Problems solved by technology

The calculation of matrix convolution is the multiplication and accumulation of a large amount of data. The traditional C / C++ language implementation requires multi-layer loop calculations, multiple memory accesses to read data, and runs on systems with traditional architectures. The calculation speed, power consumption and volume Can no longer meet the requirements of modern artificial intelligence

[0003] The existing technology generally adopts the implementation method of high-level language, which has many memory access instructions, long execution time, poor real-time performance, and low energy efficiency ratio, which cannot meet the requirements of modern image processing.

Even if the matrix convolution operation implemented by ASIC chips is relatively fragmented in design, the versatility is not strong, and the flexibility is not enough.

Images