GPU-Based Acceleration of Regression Test Suite Reduction

Abstract

If software developers adopt test automation, the costs of development and maintenance will significantly decrease and the quality of regression testing will also increase. However, the number of test cases generally grows as the software under test evolves. It will take too much time to run all of the test cases during regression testing even though test automation is adopted. This may delay the time to release software products. Thus, a test team should choose a representative set of test cases from the original test suite so that the regression testing can be accomplished in a tight build schedule and the quality of regression testing is still satisfactory. This process is called test suite reduction. The problem of test suite reduction has received considerable attention in recent decades. Many test suite reduction methods have been proposed in the literature. Yet, reducing the test suite is a time-consuming process. Performing test suite reduction is also an extra cost of regression testing. It is fortunate that General-purpose Computing on Graphics Processing Units (GPUs) are suitable to accelerate the processing of a large quantity of digital data. Thus, this paper aims to accelerate test suite reduction method using GPUs. Our empirical studies include some frequently chosen benchmarks for experimentally evaluating the effectiveness of our approach and the empirical results indicate that the presented approach works well for a test suite of high complexity.