Anzeige
Erschienen in:
2017 | OriginalPaper | Buchkapitel
Research on LogGP Based Parallel Computing Model for CPU/GPU Cluster
verfasst von : Yongwen Wu, Junqiang Song, Kaijun Ren, Xiaoyong Li
Erschienen in: Information Technology and Intelligent Transportation Systems
Aktivieren Sie unsere intelligente Suche, um passende Fachinhalte oder Patente zu finden.
Wählen Sie Textabschnitte aus um mit Künstlicher Intelligenz passenden Patente zu finden. powered by
Markieren Sie Textabschnitte, um KI-gestützt weitere passende Inhalte zu finden. powered by
Abstract
CPU/GPU heterogeneous computing has become a tendency in scientific and engineering computing. The level of heterogeneity in modern computing systems gradually rises, and CPU/GPU Heterogeneous system contains three levels of heterogeneity. Conventional parallel computation models cannot be used to estimate the running time under the CPU/GPU heterogeneous computing environment. In this paper, a new model named VLogGP is proposed, and the communication and memory access characteristics are both abstracted based on CPU/GPU heterogeneous system. We map the model to TH-1A platform, and measure all model parameters for this kind of platforms. The model can be used to study the behavior of parallel applications, estimate the execution time and guide the optimization of parallel programs.