The talk will provide an overview of recent effort towards enabling integration-based visualization on large data sets by taking advantage of modern supercomputing architectures. Integration-based visualization has garnered renewed traction in the visualization community and is a tool urgently needed by domain scientists to facilitate vector field visualization and analysis.

In this context, I will describe problems and approaches of efficient parallelization strategies that make use of distributed computation and data to achieve good performance for integral curve computation. Furthermore, I will discuss improvements to basic parallelization approaches that leverage architectural features of modern supercomputers, such as multi-core CPUs, to achieve further performance improvements. The talk concludes with an overview of open problems in large-scale vector field visualization and a brief discussion of possible future directions.