AN INTELLIGENT COMPUTING FRAMEWORK FOR MULTICORE HETEROGENEOUS EMBEDDED ARCHITECTURES

Abstract

In today’s world, advent of the multi core heterogeneous architectures finds its place in all technologies such as Internet of Things (IoT), mobile applications, health care etc. Even though these kinds of architectures have many advantages, knowledge about exploring its features for an effective usage remains in the darker side. Performance and energy consumption are two important parameters for the design of an embedded system. There is no appropriate technique or tool available for allocation of different workloads in a embedded system with multi cores for getting low energy consumption with high performance. There is no workload due to the underlying architecture, which means it can be associated with the core workload.

Many tools and test benches were designed but still the exploration in the usage of these architectures remains challenging part. To meet this problem, proposed new test bench called Selective Hybrid Intelligent Versatile (SHIV) test bench which integrates the learning algorithms to scheduling the workloads in accordance with the cores utilized and also rule engine has been formulated for the utilization of the cores in accordance with the workloads. The main aim of the test bench is to explore the usage of the embedded cores to accomplish the energy efficient with high performance. The proposed test bench has been tested with the different benchmarks such as PARSEC, COREMARK, and DHRYSTONE to evaluate its performance.