Accelerating Training and Inference of Graph Neural Networks with Fast Sampling and Pipelining