A Survey of Optimized Compiler Using Advanced Machine learning and Deep Learning Techniques
Lal Bahadur Pandey, Маниша Шарма, Rajesh Tiwari, Radhe Shyam Panda, Partha Roy · 2024
Optimizing compilers is a difficult and time-consuming task, especially when done by hand. As far as we know, the compiler handles both translation and optimization. An efficient compiler system can become more automated and simple, as evidenced by recent studies using deep learning and machine learning approaches. Model training, prediction, optimization, and feature selection are handled by most machine learning and deep learning methods. In this case, choosing the optimal characteristics is necessary in order to use deep learning and machine learning techniques to enhance the optimization quality. This study examines various approaches that might be utilized to enhance and refine the quality of the chosen heuristics as well as the general quality of machine learning and deep learning models in order to boost the compiler's efficiency. The phase-ordering problem, the amount of iterative program evaluations, and the time needed to obtain the best forecast are only a few of the many subjects covered by these approaches.