Method for solving inner product of sparse matrix and dense matrix through outer accumulation

Abstract

The invention belongs to the technical field of computer application, and particularly relates to a method for solving an inner product of a sparse matrix and a dense matrix through outer accumulation. The method comprises the steps of effective steps of firstly completing reading of original data of a sparse matrix A (MxK) and a dense matrix B (KxN); performing column-direction data compression on the sparse matrix A (MxK), and performing column-direction compression on the sparse matrix A (MxK) in the two-dimensional layout into a column-direction compressed sparse matrix A (MxK) in the one-dimensional layout; reading the outer product operation of the ith column of the column-direction compressed sparse matrix A (MxK) and the ith row of the dense matrix B (KxN) to generate ci; and accumulating the obtained outer product results to obtain the inner product of the sparse matrix A (MxK) and the dense matrix B (KxN). The original two-dimensional layout sparse matrix A (MxK) is compressed into the one-dimensional layout sparse matrix A (MxK) in the column direction, and the zero element value of the sparse matrix A (MxK) is removed, so that the calculation of invalid numerical values is avoided, the purposes of saving memory access and invalid operation are achieved, and operation efficiency is improved.

Description

technical field [0001] The invention belongs to the field of computer application technology, and in particular relates to a method for solving the inner product of a sparse matrix and a dense matrix by external accumulation. Background technique [0002] Matrix inner product multiplication is the basis of algorithms in many fields such as numerical computing, machine learning, and engineering simulation, and the inner product multiplication of sparse matrix A (MxK) and dense matrix B (KxN) is a common computing scenario. Sparse matrix A(MxK) and dense matrix B(KxN) are characterized by a certain amount of zero elements in sparse matrix A. This part of zero elements can not participate in the calculation and still ensure the correctness of the calculation results. At the same time, the storage of zero elements can be directly abandoned to improve the utilization efficiency of memory. [0003] However, in the process of calculation and storage, ignoring the zero elements in ...

AI Technical Summary

Problems solved by technology

However, in the existing outer accumulation calculation, the zero elements in the sparse matrix participate in the outer product calculation, which not only reduces the computational efficiency of the outer product operation but also increases the need to write the zero value to the memory after the outer product calculation, and then Reduced effective bandwidth for memory writes

Images