1.7. Summary

OpenCL is a new, powerful toolset for building parallel programs to run on high-performance processors. With OpenCL, you don’t have to learn device-specific languages; you can write your code once and run it on any OpenCL-compliant hardware.

Besides portability, OpenCL provides the advantages of vector processing and parallel programming. In high-performance computing, a vector is a data structure comprising multiple values of the same data type. But unlike other data structures, when a vector is operated upon, each of its values is operated upon at the same time. Parallel programming means that one application controls processing on multiple devices at once. OpenCL can send different tasks to different devices, and this is called task-parallel programming. If used effectively, vector processing and task-parallel programming provide dramatic improvements in computational performance over that of scalar, single-processor systems.