International Journal of Applied Information Systems |
Foundation of Computer Science (FCS), NY, USA |
Volume 12 - Number 36 |
Year of Publication: 2021 |
Authors: Hisham M. Alosaimi, Abdullah M. Algarni, Fathy E. Eassa |
10.5120/ijais2020451903 |
Hisham M. Alosaimi, Abdullah M. Algarni, Fathy E. Eassa . A Translation Technique for Parallelizing Sequential Code using a Single Level Model. International Journal of Applied Information Systems. 12, 36 ( March 2021), 30-40. DOI=10.5120/ijais2020451903
Running the code sequentially can be slower, and then the execution time will increase in case of the code has compute-intensive parts. Unfortunately, the sequential code does not employ the device's resources in ideal shape, because it executes one instruction at a time, which means it can perform only a single thread. To overcome the massive time taking issue while large executions, using a paralleling computing approach is a vital solution. A parallel computing code reduces the execution time by executing multiple tasks at the same time. Most researchers and programmers face some difficulties to run their sequential code as parallel due to a lack of knowledge about parallel programming models and the dependency analysis on their codes. Therefore, auto parallelization tools can be helpful to solve this issue. In this study, we have introduced a novel automatic serial to parallel code translation technique that takes serial code written in C++ as an input and generates its parallel code automatically. To validate the objectives of the current study, we compare the results of our proposed method with existing methods. Consequently, the proposed AP4OpenACC tool outperformed the other existing method mentioned in comparative analysis.