ParaView Components
The ParaView desktop application is actually a combination of three components:
- A data server that reads and processes data sets
- A rendering server that renders processed data sets into displayable graphics data
- A user interface that displays the graphics data
The ParaView Reference Manual explains the relationships between these three components and contains instructions for connecting your desktop ParaView application to remote data and/or rendering servers.
Whether in the desktop application or on a remote server, the data and rendering components of ParaView can operate in a parallel fashion, using multiple processors to complete a visualization faster. If those processors each have access to independent memory, the visualization data can be distributed among them to allow even larger data sets to be processed. In these scenarios, the ParaView client communicates directly with only one data server and one rendering server. Those servers in turn communicate with the remainder of the parallel servers of their type using the Message Passing Interface (MPI).
Setting up remote and/or parallel ParaView processes typically involves some effort, but the Texas Advanced Computing Center (TACC) has simplified the process through their Analysis Portal. This web interface allows users to log in to a node of the Stampede2 or Frontera computing cluster where they can launch an instance of the ParaView desktop application. The ParaView instance can then be connected to a set of parallel servers that both process data and perform rendering. The desktop of the login node can then be viewed on the user's own computer via a Virtual Network Computing (VNC) connection.
The following pages describe the steps that are needed to use the Analysis Portal.