Automatic Parallelization of Tiled Loop Nests with Enhanced Fine-Grained Parallelism on GPUs