AMD has posted a case study on OpenCL™ Optimization. The article discusses performance optimizations for AMD GPUs and CPUs using as a case study a simple, yet widely used computationally intensive kernel: Diagonal Sparse Matrix Vector Multiplication.
GPU Computing, a term coined by Jon Peddie, like most things, was born out of necessity when it was discovered that a cheap PC with a decent graphics card could perform engineering structural analysis for much less money than a workstation. Jem Davies, the self proclaimed god father of technical talent in ARM, does a great job of outlining what the GPU can do today, and where it's going tomorrow with OpenCL, DirectCompute and Augmented Reality.
KitGuru recently interviewed the well known and highly respected AMD employee Terry Makedon. Covering ATI Stream SDK and OpenCL, among other things, Terry gives a peak into the direct AMD is going with these technologies.
AMD announced the availability of the ATI Stream Software Development Kit (SDK) v2.1. AMD is showing their full support behind OpenCL with this update which includes support for : OpenCL / OpenGL® interoperability, OpenCL byte addressable stores, OpenCL images, OpenCL extension which support AMD media operations in OpenCL and device fission in OpenCL, along with a host of other additions. AMD also announced the launch of its own OpenCL Zone developer support website.
NVIDIA said that it could eventually port computing of physics effects created using PhysX middleware to OpenCL application programming interface (API) and capable hardware.
GPCBenchMarkOCL is a General Purpose Computing benchmark that evaluates the performance of OpenCL enabled devices with a collection of algorithms and applications. GPC benchmark can evaluate and report on the number and frequency of computing units, architecture, memory bandwidth, on-chip cache and memory and synchronize penalty.
The case study on OpenCL low poly collision detection is included in the online CMSoft OpenCL tutorials. Highly details descriptions and source code will lead you through an implementation of low polygon collision detection algorithm suitable for engineering assembly analyses. This case study uses OpenGL and Lab3D to display the 3D models.
AMD today introduced 4 new additions to its next-generation ATI FirePro family - the FirePro V7800, FirePro V5800, FirePro V4800, and FirePro V3800. Together with the the previously announced FirePro V880, AMD now offers professional graphic accelerators with OpenGL 4, OpenCL, DX11, Eyefinity multi-display support and as much as double the performance of previous models, at every price point.
Ever wonder what the difference between CUDA and OpenCL is? Streamcomputing has done a good job in clarifying the differences between CUDA and OpenCL. The article covers speed, language, heterogenous vs. homogenous and some of the terminology.
Khronos has released the OpenCL 1.0 man page XML sources. The XML source has been released under the Khronos Free Use License (BSD-like) so people can modify, repackage, redistribute etc.