Programming Massively Parallel Processors, 2nd Edition

Chapter 14

An Introduction to OpenCL^TM

Chapter Outline

14.1 Background

14.2 Data Parallelism Model

14.3 Device Architecture

14.4 Kernel Functions

14.5 Device Management and Kernel Launch

14.6 Electrostatic Potential Map in OpenCL

14.7 Summary

14.8 Exercises

References

Now that we have discussed high-performance parallel programming using CUDA C, we would like to introduce another way to exploit the parallel computing capabilities of heterogeneous computing systems with GPUs and CPUs: OpenCL^TM. In this chapter, we will give a brief overview of OpenCL for CUDA programmers. The fundamental programming model of OpenCL is so similar to CUDA that there is a one-to-one correspondence for most features. With your understanding of CUDA, you will be able ...