Decentralized Dynamic Load Distribution to Improve Speedup of Multi-Core System using Parallel Models
Computing applications that need massive computing request lead to the conflict of the obstacle regarding the development of the efficient programming models which tries to fill the gap between the hardware and software ability for applications in consideration. Therefore there is a need for research to develop an improved programming model that facilitates simple development with high performance. The proposed research work analyses various shared memory models (Pthread, OpenMP, Cilk++, TBB) in order to improve their performance parameters such as execution time and speedup of the application. In this research work the programming models with decentralized dynamic load balancing are evaluated by considering the data parallel applications and attempt has been made to verify their ability to support data parallelism with respect to the performance parameters. The motive of the proposed research work lies in incorporating virtualization with decentralized dynamic load balancing to improve the throughput of the multi-core system.
Author's Name: Rajesh Tiwari, Manisha Sharma and Kamal K Mehta