Integrating GPU APIs for maximum processing speed
Graphics Processing Units (GPU) have come a long way since their introduction in the 1970s. Initially developed purely to deliver advanced graphics to computer games, the modern GPU offers extra processing power to a vast range of technology, freeing up space on a device’s CPU and allowing today’s immense volumes of data to be transferred more quickly and efficiently.
All Active Silicon frame grabbers are compatible with GPUs and support the high levels of processing operations demanded by modern vision systems. By eliminating the reliance on a CPU, low-latency data acquisition is extremely fast as the GPU memory is made directly accessible to the frame grabber.
On paper, Intel appears to be the market leader in GPUs, but when you remove their integrated graphic solutions from the figures, NVIDIA and AMD are clear frontrunners. NVIDIA and AMD have produced APIs to support straight-forward integration with third-party devices such as frame grabbers.
NVIDIA’s GPUDirect for Video
NVIDIA introduced GPUDirect for Video® to reduce latency by transferring smaller chunks of data and avoiding unnecessary system memory copies and CPU overhead.
AMD’s DirectGMA
DirectGMA from AMD enables devices on a system bus to write directly to a portion of the GPU memory and interface with hardware to support data exchange, resulting in minimal latency data transfer and saving memory bandwidth.
GPU Solutions
To see the full advantage of installing an API and integrating GPU processing with Active Silicon’s frame grabbers, we’ve put together a fact sheet on GPU solutions. Active Silicon’s ActiveSDK includes a comprehensive suite of C++ examples for GPUDirect for Video and DirectGMA with full source code, and our support team are always on hand to assist further.
