DATA ANALYSIS FRAMEWORK FOR HETEROGENEOUS COMPUTING

Abstract

This thesis focuses on developing a data analysis framework for different computing environments. the thesis is related to emerging field of The study involves collecting and analysing data from various processors, such as AMD CPUs, NVIDIA GPUs, INTEL CPUs, and GPUs by executing various benchmarks. The primary objective is to understand the hardware-level metrics and analyse the correlation among the various performance metrics. This analysis utilizes Machine Learning and Deep Learning models for prediction and draw insights from the collected hardware-level metrics. These insights will be helpful to identify bottleneck of processors for an application.