 |
 |
 |
 |
 |
 |
 |
|
 |
 |
 |
 |
 |
netparvox, netparvox_pipe
NAME
netparvox, netparvox_pipe - Remote rendering program for regular structured
grid that interfaces with xmparvox and calls the ParVox API. netparvox runs as a standalone program
and netparvox_pipe runs in pipeline mode together with inparvox
and outparvox.
SYNOPSIS
netparvox [-server hostname] [-port N]
[-preload name from to step blk/pe idim jdim kdim dtype(0/1)]
[-prefetch frames]
netparvox_pipe [-server hostname] [-port N]
[-preload name from to step blk/pe idim jdim kdim dtype(0/1)]
[-prefetch frames]
DESCRIPTION
Netparvox and netparvox_pipe are the remote renderer for
ParVox that
runs on the parallel
machine. It interfaces with the GUI xmparvox
over a Unix TCP socket. Netparvox is a standalone renderer for regular structured
grid. Netparvox_pipe runs together with inparvox and netparvox.
- -server hostname
- When specified, this option will cause netparvox to connect to
xmparvox running on hostname. If this option is
not specified, then netparvox will act as a server and
wait for an incoming connection by xmparvox.
- -port N
- Specifies the port number to use to connect to xmparvox.
If this is set, then xmparvox should also set this parameter.
The default port is 6543.
- -preload name from to step blk/pe idim jdim kdim dtype(0/1)
- Preloads data before starting the connection. Name
specifies the filename of the data to load, using the ``name####''
format for multiple timestep files, using one `#' for each
fixed digit position. From, to, and step
are the starting, stopping, and the interval of the file numbers to
load. The blk/pe and the idim jdim kdim specifies
how the data is read in and distributed over the nodes. If
blk/pe is positive, then idim jdim kdim will be
ignored. Finally, the dtype specifies the input data file
type. 0 is the raw format, which isn't being officially
supported. 1 is the NetCDF format.
- -prefetch N
- Used for out-of-core rendering; only N timesteps will be loaded into memory at a time.
New timesteps will be loaded in during animation.
To run netparvox on a parallel machine, use
% mpirun -np n netparvox
To run netparvox_pipe on SGI machines, you have to prepare a file, say mpi.file including the following lines:
-np n1 netparvox_pipe :
-np n2 outparvox :
-np n3 inparvox
then run
% mpirun -f mpi.file
The above command will start netparvox_pipe on n1 processors, outparvox on n2 processors, and
inparvox on n3 processors. Note you have to run outparvox on at least 3 processors.
Netparvox and xmparvox must be
started in the proper order. The first
process to start should not use the -server hostname option.
The second process must then use the -server
hostname to connect to the first process.
SEE ALSO
netparvox_pipe_ug,
inparvox, outparvox
xmparvox,
ParVox Tutorial: Using netparvox
|
 |
 |
 |
 |
 |
|
 |
 |
 |
 |
 |
 |
|
 |
This page,
http://pat.jpl.nasa.gov/public/ParVox/manpages/netparvox.html,
is maintained by
and was last modified Tuesday, 22-Mar-2005 16:23:59 PST
|
|
 |
 |
 |
 |
 |
 |
 |
|