by Ilya Gavrichenkov
08/22/2008 | 11:43 AM
Nvidia today released CUDA 2.0, a new version of its C language development environment for its graphics processing units (GPUs). CUDA enables PCs equipped with certain Nvidia graphics processors to offload the computation of some information to the graphics processor instead of using the CPU. Nvidia's graphics processors are parallel processing engines, which makes them suited to handling tasks such as medical volume reconstruction and oil and gas seismic computing, for example.
CUDA is Nvidia's implementation of technology called General-Purpose Computing on Graphics Processing Units, or GPGPU. CUDA was developed by NVIDIA in 2006 as an architecture for advanced programming that takes advantage of additional capabilities offered by parallel processing with multiple cores. This compiler works by feeding calculation data to the massively multi-cored GPU systems, allowing them to break down the code in a shorter span of time than is normally required by sequentially-operating main processors.
NVIDIA claims there are 90 million CUDA compatible GPUs already purchased by customers around the world. Apps written with CUDA require a card with one of NVIDIA's chips built with its unified shader architecture, including all GeForce 8- and 9-series cards as well as newer Quadro FX cards. Its Tesla and Quadro Plex general-purpose acceleration cards also support the feature.
The latest production release of the CUDA toolset includes Windows 32-bit and 64-bit support as well as Mac OS X support, and also includes an Adobe Photoshop plug-in example, as well as its source code. The example allows developers to design plug-ins that move the most compute-intensive functions of Adobe Photoshop to the GPU, such as filtering and image manipulation, delivering performance improvements. The plug-in is available as source code so developers can easily develop advanced filters and imaging techniques that are available directly within Adobe Photoshop. Nvidia hopes the example will spawn further efforts from third-party developers to create filtering and image manipulation software that leverages Nvidia GPUs to speed processing.