NVIDIA's GPUDirect For Video Technology Loves That Parallel Processing
The company claims that their technology is the fastest, most deterministic way to get video to the GPU for processing, by enabling leading video I/O board manufacturers to take advantage of the programmability of the GPU. With GPUDirect for Video, the video I/O card is fully synchronized with the GPUs, eliminating CPU overhead and, most importantly, dramatically reducing latency.
Historically, broadcasters and video production professionals had to contend with delays of as many as ten frames per second -- an amount easily visible to the eye -- when transferring video from a video I/O device to a GPU. This is because the two devices previously could not communicate directly with one another to ensure synchronization. This required application developers to manage complex buffering schemes, leading to unnecessary CPU overhead and increased latency. Most importantly, these delays in aggregate significantly raised the cost and complexity of larger broadcast productions. Might sound a little bit outside of the consumer space, but then again, you may well notice the benefits of the technology in your favorite TV program in the not so distant future.
NVIDIA GPUDirect for Video technology is available immediately, free of charge, directly from NVIDIA for video I/O manufacturers, proprietary hardware providers, and video switcher manufacturers.