With DSPs, GPUs, and FPGAs a wide range of acceleration technologies for embedded systems exists today. They solve even computationally intense tasks quickly and energy efficiently. Using the latest integration of ARM technology and adaptable high-speed FPGA fabric in Xilinx ZYNQ a perfect platform for high performance signal and image processing for many domains is met.
Targeting the design for all these acceleration platforms, CoSynth presents an OpenCL based runtime environment. The OpenCL support delivers a universal interface to whatever accelerator architecture is available. The application is usable on and optimizable for any hardware. Porting to another platform requires only a retargeting of the computation kernels onto the accelerator architecture.
Within the EU funded project FiPS CoSynth provides technologies with hardware and software modules for the development of customer specific systems based on standard (COTS) and newly developed components.
At the booth CoSynth presents image processing with Xilinx ZYNQ and Linux, combining hardware accelerated image preparation and software based image processing with OpenCL. The demonstrator features CameraLink, Xilinx ZYNQ 7020, and full Linux support including the OpenCL framework for access to the FPGA kernels.