Hello, I am new in OpenCL and i have a simple question.
For my program, i need firts to subdivise my data (image 2D data) into sevral parts and load all parts in to the GPU memory. After that i need to run the kernel on this differents subdivisions.
There is an overlapping with each subdivions and i don’t know how to associate the kernel with a part of the image.
Why do you need to subdivide the actual data? Can you get away with just passing the entire image to each kernel, even though each kernel launch runs on a part of the image? Then, if you want to associate a particular kernel launch with a region of the image, you could just pass in the region paramaters (x, y, width, height) as input arguments to the kernel. If the data needed for the computation overlaps with data needed for other kernel launches, it should be no big deal, since each kernel launch has access to the whole image.