Nvidia is rolling out the latest version of its CUDA Toolkit and its software development kit, which includes support for Microsoft’s upcoming Windows 7 operating system.
The May 8 release of CUDA Toolkit 2.2 and SDK was the latest in a string of product announcements made the same week the graphics chip maker said it had lost $201.3 million in the first quarter.
The tool kit, SDK and drivers can be downloaded here.
CUDA is the parallel computing engine for Nvidia’s GPUs. In addition to the support for Windows 7-which Nvidia officials said supports computing using GPUs-CUDA Toolkit 2.2 also includes the CUDA Visual Profiler for profiling C applications running on the GPU, improved performance for OpenGL applications, such as medical imaging, and twice the bandwidth savings for video processing applications.
The new version also brings performance improvements for streaming media, video transcoding, image processing and signal processing applications by enabling CUDA functions to read and write directly from pinned system memory. That reduces the amount of data that needs to be copied between the GPU and CPU.
Applications that use multiple GPUs will have better performance and use less memory because multiple GPUs will be able to access the same data in system memory. Application performance is improved by having memory copied asynchronously, and the CUDA Toolkit now includes a hardware debugger for the GPU.
The release of CUDA Toolkit 2.2 comes shortly after Nvidia’s announcement that it is teaming with Dell to offer Dell workstations with Nvidia’s Tesla GPU technology and that it is rolling out a preconfigured GPU cluster.
Nvidia announced May 7 that it saw revenues in the first quarter drop 42 percent, to $664.2 million, from $1.15 billion during the same period in 2008.