Chapter 10. Advanced OpenACC
Jeff Larkin, NVIDIA
With the basics of OpenACC programming well in hand, this chapter discusses two advanced OpenACC features for maximizing application performance. The first feature is asynchronous operations, which allow multiple things to happen at the same time to better utilize the available system resources, such as a GPU, a CPU, and the PCIe connection in between. The second feature is support for multiple accelerator devices. The chapter discusses two ways that an application can utilize two or more accelerator devices to increase performance: one using purely OpenACC, and the other combining OpenACC with the Message Passing Interface (MPI).
10.1 Asynchronous Operations
Programming is often taught by developing ...
Get OpenACC for Programmers: Concepts and Strategies, First Edition now with the O’Reilly learning platform.
O’Reilly members experience books, live events, courses curated by job role, and more from O’Reilly and nearly 200 top publishers.
