Abstract
The development of parallel solutions over contemporary heterogeneous supercomputers is complex and challenging, especially for coding, performance analysis, and behavioral characterization. The task-based programming model is a possible alternative to adequately reduce the burden on the programmer. Such model consists of dividing the application into tasks with dependencies through a directed acyclic graph (DAG), and subject the DAG to a runtime scheduler that will map tasks to resources. In this paper, we present the design, development, and performance analysis of a task-based heterogeneous (CPU and GPU) application of a Computational Fluid Dynamics (CFD) problem that simulates the flow of an incompressible Newtonian fluid with constant viscosity. We implement our solution based on the StarPU runtime and use the StarVZ toolkit to conduct a comprehensive performance analysis. Results indicate that our solution provides a 6.5\(\times \) speedup compared to the serial version on the target machine using 7 CPU workers and a 60\(\times \) speedup using 5 CPU and 2 GPU workers.
Sign-in/Register to access full text options 
Talk to us
Join us for a 30 min session where you can share your feedback and ask us any queries you have
Schedule a callDisclaimer: All third-party content on this website/platform is and will remain the property of their respective owners and is provided on "as is" basis without any warranties, express or implied. Use of third-party content does not indicate any affiliation, sponsorship with or endorsement by them. Any references to third-party content is to identify the corresponding services and shall be considered fair use under The CopyrightLaw.
