Sparse matrix storage revisited
Malik Silva
Abstract
Malik Silva
Abstract
In this paper, we consider alternate ways of storing a sparse matrix and their effect on computational speed. They involve keeping both the indices and the non-zero elements in the sparse matrix in a single data structure. These schemes thus help reduce memory system misses that occur when the usual indexing based storage schemes are used to store sparse matrices and give promising performance improvements.
OpenAlex reports 10 citations for this work. Citation counts describe recorded attention and do not establish research quality.
A contribution statement is not available in the OpenAlex record.
Method details are not available in the OpenAlex metadata.
Findings are not separately available in the OpenAlex metadata.
Limitations are not available in the OpenAlex metadata.
Application details are not available in the OpenAlex metadata.
In this paper, we consider alternate ways of storing a sparse matrix and their effect on computational speed. They involve keeping both the indices and the non-zero elements in the sparse matrix in a single data structure. These schemes thus help reduce memory system misses that occur when the usual indexing based storage schemes are used to store sparse matrices and give promising performance improvements.
Key concepts: Sparse matrix, Computer science, Matrix (chemical analysis), Search engine indexing, Computer data storage, Matrix-free methods, Storage efficiency, Parallel computing