svds-C: A Multi-Thread C Code for Computing Truncated Singular Value Decomposition

TL;DR

svds-C is a high-performance, multi-threaded C implementation for truncated SVD that significantly outperforms existing tools in speed and memory efficiency while maintaining accuracy.

Contribution

This paper introduces svds-C, a parallel C program for truncated SVD that improves speed and memory usage over previous implementations and state-of-the-art algorithms.

Findings

svds-C runs 4.7X to 12X faster than svds with 16 threads.

svds-C uses about half the memory of svds.

svds-C outperforms other algorithms in speed and robustness.

Abstract

This article presents svds-C, an open-source and high-performance C program for accurately and robustly computing truncated SVD, e.g. computing several largest singular values and corresponding singular vectors. We have re-implemented the algorithm of svds in Matlab in C based on MKL or OpenBLAS and multi-thread computing to obtain the parallel program named svds-C. svds-C running on shared-memory computer consumes less time and memory than svds thanks to careful implementation of multi-thread parallelization and memory management. Numerical experiments on different test cases which are synthetically generated or directly from real world datasets show that, svds-C runs remarkably faster than svds with averagely 4.7X and at most 12X speedup for 16-thread parallel computing on a computer with Intel CPU, while preserving same accuracy and consuming about half memory space. Experimental results also demonstrate that svds-C has similar advantages over svds on the computer with AMD CPU, and outperforms other state-of-the-art algorithms for truncated SVD on computing time and robustness.