A Map-Reduce Based Framework for Heterogeneous Processing Element Cluster Environments
CCGRID '12 Proceedings of the 2012 12th IEEE/ACM International Symposium on Cluster, Cloud and Grid Computing (ccgrid 2012)
EG PGV'10 Proceedings of the 10th Eurographics conference on Parallel Graphics and Visualization
A Hybrid Resource Reservation Method for Workflows in Clouds
International Journal of Grid and High Performance Computing
A Hybrid Resource Reservation Method for Workflows in Clouds
International Journal of Grid and High Performance Computing
Hi-index | 0.00 |
With the rapid development of GPU (Graphics Processor Unit) in recent years, GPGPU (General-Purpose computation on GPU) has become an important technique in scientific research. However GPU might well be seen more as a cooperator than a rival to CPU. Therefore, we focus on exploiting the power of CPU and GPU in solving generic problems based on collaborative and heterogeneous computing environment. In this work we present a parallel processing paradigm based on CPU-GPU collaborative computing model to optimize the performance of task scheduling. In addition, we evaluate a new task scheduling algorithm using NVIDIA GeForce 7600GT compare with traditional task scheduling algorithm. The results show that our algorithm increase average performance of 26.5% compared with traditional algorithm. Based on our results and current trends in microarchitecture, we believe that efficient use of CPU-GPU collaborative environment will become increasingly important to high-performance computing.