Khronos publishes definitive OpenCL 2.0 standard

The Khronos Group, responsible for the OpenGL standards, among other things, has approved the 2.0 specification for OpenCL. OpenCL 2.0 includes improved memory access and new image manipulation capabilities.

OpenCL is both a programming language based on C++11 and C11 and a platform. The open standard, which is supported by Apple, Intel, Nvidia, Qualcomm and AMD, among others, is intended for parallel calculations on cpu and gpu, also known as gpgpu. The OpenCL 2.0 standard must follow the two-year-old 1.2 specification. After a draft was published by the Kronos Group in July, a final standard has been established following third-party input lined up.

The OpenCL 2.0 standard contains several new possibilities for applications that want to use the computing power of gpgpu. For example, the host and device kernels can now share the virtual working memory with each other, so that speed gains can be achieved. Also, programmers can now use nested parallel calculations and there is more freedom in formulating functions.

For image manipulation, OpenCL 2.0 provides support for srgb and 3d images. It is also possible to write and read data from an image at the same time, and images can be generated based on OpenGL textures. There is also support for the Android operating system.