NVIDIA has released their CUDA Toolkit 3.2. Lots of new goodness in this version, with special note the new OpenCL support. This means you can now use one toolkit for both CUDA and OpenCL. Support is currently only for Linux and Windows.
AMD to hand out free copies of their newly-published book "OpenCL programming on AMD CPU/GPUs" at the Khronos DevU on December 8th, and the December 9th Conference and Lunch.
AMD has posted a great Google Map showing Universities world-wide that offer OpenCL coursework.
The OpenCL supported Blender Compositor shows significant gains on OpenCL capable machines.
NVIDIA has brought the Fermi graphics architecture to the Mac with the launch of the Quadro 4000 for Mac. The workstation-class video supports the same features as its Windows counterpart and focuses heavily on general-purpose computing. OpenCL in Snow Leopard also gets a boost from this release and can greatly accelerate apps that are using NVIDIA's own CUDA language as well, such as video processing in Adobe Premiere Pro CS5.
Available at WhatIf.intel.com, this implementation delivers OpenCL 1.1 specification features optimized for Intel Core™ processors for developers desiring to explore CPU advantages found on many OpenCL workloads. Currently, Microsoft Windows 7 and Windows Vista operating systems are supported. Intel OpenCL SDK is a full implementation of the OpenCL 1.1 specification, including all API's and language features, and supports additional optional features and extensions such as: Out-of-order Execution model, Images support, Double precision floating point, OpenCL-OpenGL interoperability, and more. This SDK is not fully conformant yet.
AMD will be hosting a bunch of cool demo's at their booth (#3119). Demo's range from the latest 12-core AMD OpteronTM 6100 Series processor-based OEM servers to some new applications leveraging AMD graphics technology and OpenCL™.
The Khronos Group congratulates ARM on the announcement of the Mali-T604 and its plan to support full profile OpenCL 1.1 on both ARMv7 CPUs and the GPU.
Researchers from the University of Warwick’s Performance Computing and Visualization Department and Oxford University’s eResearch Centre have put together a study: should we install a GPGPU-based system or a more traditional IMB Blue Gene-like supercomputer? The team’s research will present their work at the 1st International Workshop on Performance Modeling, Benchmarking and Simulation of High Performance Computing Systems at the SC10 conference in New Orleans. The Khronos Group will be at Booth # 1132 at SC10.
Currently there are several ways to feed data to the GPU no matter of what API we use and what type of application we develop. In case of OpenGL we have uniform buffers, texture buffers, texture images, etc. The same is true for OpenCL and other compute APIs that even provide more fine-grained memory management taking advantage of the local data store (LDS) available on today’s hardware. In this article I’ll present the memory access performance characteristics of AMD’s Evergreen-class GPUs focusing on what this all means from OpenGL point of view. While most of the data is about the HD5870, the general principles and relative performance characteristics are valid for other GPUs, including ones from other vendors.