The Khronos Group announced the finalization of the OpenCL 2.0 standard. Among other things, this new specification delivers support for shared virtual memory, nested parallelism, generic address space and more.
The Khronos™ Group today announced the ratification and public release of the finalized OpenCL™ 2.0 specification. OpenCL 2.0 is a significant evolution of the open, royalty-free standard that simplifies cross-platform, parallel programming. With an enhanced execution model and a subset of the C11 and C++11 memory model, synchronization and atomic operations, OpenCL now enables a significantly richer range of algorithms and programming patterns to be easily accelerated with improved performance. Significant feedback from the developer community was incorporated into the final specification, following its provisional release in July. The OpenCL 2.0 specifications are available at www.khronos.org/opencl/.
“Khronos received significant and thoughtful developer feedback from the provisional release of OpenCL 2.0, much of which has been adopted, or will be merged with emerging hardware capabilities as this state-of–the-art parallel programming platform continues to evolve,” said Neil Trevett, chair of the OpenCL working group, president of the Khronos Group and vice president of mobile content at NVIDIA. “OpenCL continues to gather momentum on desktop, mobile and embedded devices, including providing a unified programming environment for dynamically balancing diverse CPU, GPU, DSP and hardware resources in mobile SOCs for advanced use cases ranging from vision processing for Augmented Reality to physics simulation for mobile gaming.”
OpenCL 2.0 updates and additions include:
Shared Virtual Memory
Host and device kernels can directly share complex, pointer-containing data structures such as trees and linked lists, providing significant programming flexibility and eliminating costly data transfers between host and devices.
Device kernels can enqueue kernels to the same device with no host interaction, enabling flexible work scheduling paradigms and avoiding the need to transfer execution control and data between the device and host, often significantly offloading host processor bottlenecks.
Generic Address Space
Functions can be written without specifying a named address space for arguments, especially useful for those arguments that are declared to be a pointer to a type, eliminating the need for multiple functions to be written for each named address space used in an application.
Improved image support including sRGB images and 3D image writes, the ability for kernels to read from and write to the same image, and the creation of OpenCL images from a mip-mapped or a multi-sampled OpenGL® texture for improved OpenGL interop.
A subset of C11 atomics and synchronization operations to enable assignments in one work-item to be visible to other work-items in a work-group, across work-groups executing on a device or for sharing data between the OpenCL device and host.
Pipes are memory objects that store data organized as a FIFO and OpenCL 2.0 provides built-in functions for kernels to read from or write to a pipe, providing straightforward programming of pipe data structures that can be highly optimized by OpenCL implementers.
Android Installable Client Driver Extension
Enables OpenCL implementations to be discovered and loaded as a shared object on Android systems.