Fluent Quick Start
For tutorials and interactive use (eg: creating a model) you should run fluent on p3n14-c. To access this node, use the following steps:
- First, make sure you have an X server such as Cygwin running on your desktop PC. Cygwin has SSH (secure shell) built-in, but if you're using a different package such as Hummingbird then you may also need an SSH program such as Putty. Ask your lab technician if you need help with setting up an X server and SSH connection.
Use SSH to log in to kerr.canterbury.ac.nz - make sure you are forwarding your X server information. Check this by running
after you log in. You should see output similar to localhost:16.0 from the command. No output or an error is bad.
To run, say, the 2d version of Fluent, use this command:
All going well, the Fluent graphical interface will show on your desktop display after 30 seconds or so.
You can run Fluent interactively for very small jobs, like tutorials, and for interactive tasks like tweaking a mesh. However, for any real computation you must use Fluent via a script on the batch queue system. For an example Fluent script, see our LoadLeveler page.
Once you've got Fluent started, you can access some excellent online tutorials to learn how to use it, such as
- The Official Fluent Tutorials
- The Cornell University Tutorials
- ... or run a Google search for "Fluent Tutorial"
Paging Error - out of memory
A very large model of 5 million elements with 7 equations per element proved to be too large to run on the development node - it would run for a little while, then give a paging (out of memory) error. The solution is to run the model on a larger set of processors say: four CPUS on each of four nodes seems fine, possibly 16 CPUs on one node would work as well, but if memory is a problem, use more than one node.
Run /usr/local/bin/harpoon on p1n14.