my code is work on opencl and d3d9 interop,
in kernel use write_imagef is cost 4ms to write image data?
the code:
AllVal = (mul24(val, iInvWeight ) + mul24(v2Val, iWeight))>>11;
write_imagef(image, id, convert_float4(AllVal )/(float4)255.0);
how do i to optimize my code?thank you !!