With Blender 2.79, OpenCL support has improved and should be closer to parity with Blender's CUDA capabilities. The OpenCL Cycles renderer has shorter render times by up to 50% in some cases, tiles are now seen updating while rendering, support for SSS and volume rendering, optimized transparent shadows, and various fixes.
The Portable Computing Language (POCL) has issued a new release of their open-source CPU-based OpenCL implementation. This new version of POCL continues relying upon LLVM and with this release adds support for LLVM/Clang 4.0 and 3.9.
The GPU Technology Conference (GTC2017) will be running from May 8-11 this year in San Jose Convention Center. This year will see many sessions related to Khronos Technology including OpenCL, OpenGL, OpenVX, Vulkan and WebGL. Check a list of Khronos technology only sessions on the Khronos site, or visit the NVIDIA GTC site to see all sessions.
Xilinx, Inc announced expansion into a wide range of vision guided machine learning applications with the Xilinx reVISION stack. Developers with limited hardware expertise can use a C/C++/OpenCL development flow with industry-standard frameworks and libraries like Caffe and OpenCV to develop embedded vision applications on a single Zynq SoC or MPSoC. For application level development, Xilinx supports industry-standard frameworks including Caffe for machine learning and OpenVX for computer vision.
Furian is designed to address the increasing compute requirements across multiple applications and market segments with efficient use of compute APIs including OpenCL 2.0, Vulkan 1.0 and OpenVX 1.1*. Furian adds a bi-directional GPU/CPU coherent interface for efficient sharing of data; and a transition to user mode queues from kernel mode queues which reduces latency and CPU utilization for compute operations. Based on a published Khronos specification, GPUs based on the PowerVR Furian architecture are expected to pass the Khronos Conformance Testing Process. Current conformance status can be found at www.khronos.org/conformance.
This project is an OpenCL-based simulator for brain models built using Nengo. It can be orders of magnitude faster than the reference simulator in nengo for large models. Nengo is a Python library for creating and simulating large-scale brain models.
NVIDIA graphics driver for Windows version 378.66 is now offering some OpenCL 2.0 support. From the release notes: "New features in OpenCL 2.0 are available in the driver for evaluation purposes only." Some known issues include: The current implementation is limited to 64-bit platforms only; OpenCL 2.0 allows kernels to be enqueued with global_work_size larger than the compute capability of the NVIDIA GPU. The current implementation supports only combinations of global_work_size and local_work_size that are within the compute capability of the NVIDIA GPU; For executing kernels (whether from the host or the device), OpenCL 2.0 supports non-uniform ND-ranges where global_work_size does not need to be divisible by the local_work_size. This capability is not yet supported in the NVIDIA driver, and therefore not supported for device side kernel enqueues.