Yesterday, Nvidia announced they would release their OpenCL driver and software development kit (SDK) is being offered to developers participating in their OpenCL Early Access Program. The beta release has been designed to obtain feedback from early-adoption developers about the implementation, its performance, functionality, stability and compatibility.
##CONTINUE##
Tony Tamasi, Nvidia’s Senior VP of technology and content said, “The OpenCL standard was developed on Nvidia GPUs and [we were] the first company to demonstrate OpenCL code running on a GPU. Being the first to release an OpenCL driver to developers cements [our] leadership in GPU Computing and is another key milestone in our ongoing strategy to make the GPU the soul of the modern PC.”
OpenCL is short for Open Computing Language and describes a technique for writing software which allows software to be written for disparate heterogeneous architectures, such as an x86 CPU and a Nvidia or ATI GPU. It is a standard maintained by a non-profit organization called Khronos Group.
The OpenCL language/interface allows code to be developed within a single application which can then execute simultaneously on multiple architectures within the target machine. Such an encapsulating language allows Windows or Linux developers, for example, to not only write code for their target application on the CPU and OS in question, but also in a generic way for execution on the video card’s GPU itself.
Whereas utilization of the GPU is possible today using Nvidia’s CUDA or ATI’s CTM/BrookGPU implementations, having an OpenCL framework is, in theory, something akin to OpenGL or OpenAL, which are open 3D graphics and audio standards enabling developers to write code for multiple platforms which support the standard in drivers. The difference here is that the code developed using OpenCL operates directly on the GPU it’s targeting with the graphics card driver operating differently than it does with CUDA or CTM/BrookGPU API function calls. In theory, an OpenCL application would be “more native” and should operate faster.
The OpenCL standard was originally developed by Apple. As such, it will be supported natively in the next Mac OS X 10.6 release, dubbed Snow Leopard. Such support will enable greatly accelerated applications to be created for the Intel-architecture Mac devices, those which enable existing video and audio crunching software to move much faster, for example. In addition, AMD has also reportedly abandoned its CTM technology in favor of supporting DirectX 11 and OpenCL as well.
Basically, OpenCL simplifies everything related to writing code across multiple heterogeneous processors (CPU and GPU, and theoretically others as well).
See Nvidia’s press release. Developers can apply to the OpenCL Early Access Program here.
-----------------------------
BY Rick Hodgin
Source：geek.com
© 1996-2009 Geek.com, LLC - All rights reserved.
Subscribe to:
Post Comments (Atom)
Posts
0 comments:
Post a Comment