CNN-oriented batch matrix multiplication parallel optimization method and system on SW architecture

By converting the input feature map and convolution kernel of the CNN convolutional layer into a matrix on the Shenwei architecture, combining DMA transmission and dynamic row block division algorithms, the batch matrix multiplication task is optimized, which solves the communication bottleneck and load imbalance problems of the Shenwei platform and improves the computing performance.

CN120508740AActive Publication Date: 2025-08-19QILU UNIVERSITY OF TECHNOLOGY (SHANDONG ACADEMY OF SCIENCES) +1
5 Cites 2 Cited by

Patent Information

Application Number
CN202511020934.6
Authority / Receiving Office
CN · China
Patent Type
Applications(China)
Current Assignee / Owner
Filing Date
2025-07-24
Publication Date
2025-08-19
Estimated Expiration
2045-07-24

AI Technical Summary

Technical Problem

The existing technology lacks adaptive batch matrix optimization solutions on domestic Shenwei supercomputer platforms, and there are problems of communication bottlenecks and load imbalance, resulting in limited computing performance, especially in high concurrency and fine-grained computing scenarios.

Method used

The main and slave cores are designed in a collaborative manner, and the input feature map and convolution kernel of the CNN convolution layer are converted into input matrix and weight matrix. The parameter structure array is transmitted through a single DMA, combined with dynamic row block division algorithm and double-buffered DMA asynchronous prefetching, and parallel processing of matrix multiplication tasks is optimized.

Benefits of technology

It improves the parallel processing efficiency between the master and slave cores of Shenwei processors, significantly improves computing performance, especially in high concurrency and fine-grained computing, and reduces communication latency and data access latency.

✦ Generated by Eureka AI based on patent content.
Patent Text Reader

Abstract

The invention provides a CNN-oriented batch matrix multiplication parallel optimization method and system on a SW architecture, and belongs to the technical field of artificial intelligence parallel optimization. Comprising the following steps: respectively converting an input feature map and a convolution kernel in a convolution layer into an input matrix and a weight matrix, and processing the input matrix and the weight matrix into a plurality of groups of independent matrix multiplication tasks in batches; the main core encapsulates a matrix multiplication task into a parameter structure array, the parameter structure array is transmitted to the slave core through single DMA, and the slave core divides rows of an input matrix into row block tasks by adopting a dynamic row block division algorithm according to the total number of threads and the height of the matrix; and executing sub-matrix multiplication calculation on the distributed independent row blocks, asynchronously prefetching matrix sub-blocks by adopting a double-buffer DMA (Direct Memory Access), and executing matrix multiply-accumulate calculation. The parallel processing efficiency of batch matrix multiplication between the master core and the slave core of the SW processor can be improved, and the algorithm performance is optimized.
Need to check novelty before this filing date? Find Prior Art