Performance Comparison of Open Source and Commercial Computing Tools in Educational andOther Use — Scilab vs. MATLAB

Abstract

In this paper, the authors compare the features and the overall performance of the two high-level numerical computing and modeling software environments: the freeware Scilab and commercially available industry-standard MATLAB. The motivation for the work emanated from the educational use of these tools at the college and university level, but with a perspective to their professional and scientific use as well. Their performance is tested by measuring the execution times of several combined-task benchmarks implemented as test functions, built upon nine common numerical tasks that are often found in programs for solving standard engineering problems. They include basic algebra and matrix calculations, signal generation, signal analysis, and storing and retrieving data to and from the hard disk drive. Although MATLAB outperforms Scilab in all the benchmarks except the disk file manipulations, in the presumed vectorization versions of the benchmarks, it is not for much. The overall performance of the freeware rival is very satisfactory, making it a good choice not only for educational use but also for scientific and professional purposes, especially when funding is critical.