NEW APPROACH TO PREDICTION OF MEMORY LEAK IN HPC HIGH-PERFORMANCE COMPUTING BY USING MPI (MESSAGE PASSING INTERFACE)

Abstract

The analysis has been done to describe how the memory leaks in HPC system can be done better that has been utilised by MPI. The beginning of this work has been done by defining a introduction which illustrates the significance of the memory leak analysis in HPC. the entire work started with this segment and in identifying this challenge, it also illustrates the Artificial Intelligence and the interface known as the message passing interface. However, the entire framework illustrates diagnosis and anomalies, detailing memory leak prediction approaches and techniques. Hence, the main focus of this method has been to enhance MPI based HPC memory leak analysis. Shipping in containers and algorithmic forecasting are used to achieve this aim. The approach covers MPI data collection and Machine Learning model development in detail. The findings and analysis imply that Decision Trees and Random Forests may efficiently discover abnormalities in many HPC systems. Success with these strategies supports this. In determination, choice of features, model versatility, and multidisciplinary cooperation are crucial for boosting the leakage of memory estimation in supercomputer mechanisms