Chapter 20
CUDA Dynamic Parallelism
Chapter Outline
20.1 Background
20.2 Dynamic Parallelism Overview
20.3 Important Details
20.4 Memory Visibility
20.5 A Simple Example
20.6 Runtime Limitations
20.7 A More Complex Example
20.8 Summary
CUDA dynamic parallelism is an extension to the CUDA programming model enabling a CUDA kernel to create new thread grids by launching new kernels. Dynamic parallelism is introduced with the Kepler architecture, first appearing in the GK110 chip. In previous CUDA systems, kernels can only be launched from the host code. Algorithms that involved recursion, irregular loop structures, time-space variation, or other constructs that do not fit a flat, single level of parallelism needed to be implemented with ...
Get Programming Massively Parallel Processors, 2nd 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.
