Effective visualization of 3D flow field plays an important role in understanding underlying features presented in the flow field data generated from simulations. Streamline and stream surface are two standard way to visualize flow field. Streamline is a curve traced from a particle inside the flow field while stream surface is the union of streamlines which traced from seeds that are densely sampled on a seeding curve. Nowadays, as the data generated from simulations become larger and larger, parallel generation of streamlines and stream surfaces become crucial for visualizing flow field. In this talk, I will talk about the project we are working on to parallel generate stream surface on distributed memory machine.  I will cover the basic stream surface algorithm, our strategy to parallel the stream surface algorithm and some initial benchmarking results. At last, I will talk about the future work such as the load balancing.