I have to execute some processing for a set of points on an image, let say an image of 1024x768... it means 786432 points.
Is there a way to execute an OpenCL kernel on the 786432 points in one step without a crash ?
I have try but for now, OpenCL crash 😞
What solution do I have ?
Thanks
Are you sure memory limitation is not the reason for crash? It should be possible if memory is not an issue. Could you please try to run on CPU?
Hi,
I run it on the CPU only (I have no GPU... hum an Intel graphic card !)
When I execute by chunk of 1024 (Work group size) it run correctly !
Thanks
which work group size it is crashing? could you please give us test case which shows crash?