Exploiting Parallelism in Iterative Irregular Maxflow Computations on GPU Accelerators