I have an openCL kernel that is compiling for an intel processor but not the GPU. I am using doubles in the kernel. What is the method for enabling doubles in an openCL kernel. I believe this is what is causing the failed compilation of the gpu image because when I change everthing to floats it compiles fine(The GPU version).
do your GPU support cl_khr_fp64 extension? you can find out with clinfo program. and add at the start of the kernel #pragma OPENCL EXTENSION cl_khr_fp64 : enable