If you generate quadrillions of data points in a simulation study, looking at the data by rendering it on your workstation can be problematic: It takes days to rotate the image by one degree.
To speed up the process, scientists at Argonne National Laboratory are exploring software-based parallel volume rendering performed on the 160,000-core Blue Gene/P supercomputer. They've scaled up to problem sizes of over 80 billion voxels per time step and generated images up to 16 megapixels. There's more information about the project on Argonne's news site.
The image below shows the rendering of a simulated core-collapse supernova created with this technique.