The API below is used to launch a kernel. Initialize work_dim, global_work_size, and local_work_size so that a 3-dimensial kernel is launched. Assume that the total number of threads is 1,000,000. The global number of threads per dimension is the same and there are 5 work-groups per dimension. Ignore the other arguments of clEnqueueNDRangeKernel().

cl_int clEnqueueNDRangeKernel ( cl_command_queue command_queue,

cl_kernel kernel,

cl_uint work_dim,

const size_t *global_work_offset,

const size_t *global_work_size,

const size_t *local_work_size,

cl_uint num_events_in_wait_list,

const cl_event *event_wait_list,

cl_event *event)