everty time compiling kernel in AMD and Nvidia speed is acceptable but in Intel it quite slow (my GPU Intel HD4000 compile cl file with 600 line it take over 2 minutes)
So we can save compiled kernel with clGetProgramInfo( _program, CL_PROGRAM_BINARIES, sizeof( unsigned char* ) * nDevices, pgBinaries, NULL ); and next time loading clCreateProgramWithBinary
everty time compiling kernel in AMD and Nvidia speed is acceptable but in Intel it quite slow (my GPU Intel HD4000 compile cl file with 600 line it take over 2 minutes) So we can save compiled kernel with clGetProgramInfo( _program, CL_PROGRAM_BINARIES, sizeof( unsigned char* ) * nDevices, pgBinaries, NULL ); and next time loading clCreateProgramWithBinary