CSM Software - Frequently Asked Questions
1. How do I get started with Abaqus at NCSA ?
Abaqus is open to all academic users at NCSA. Users with valid allocations
must be placed into the Abaqus access list. Please mail your requests with a brief description
of your project to csm@ncsa.uiuc.edu. All new users
are strongly encouraged to visit NCSA's Abaqus and High-Performance Computing Environment pages for quick start-up information.
Send e-mail to csm@ncsa.uiuc.edu for all Abaqus
related questions.
Commercial use of Abaqus is governed by the NCSA Private Sector Program.
Please contact skoric@ncsa.uiuc.edu for more information.
2. How do I get started with ANSYS at NCSA? Which Ansys License
?
ANSYS is open to all academic users at NCSA. Users with valid allocations
must be placed into the ANSYS access list. Please mail your requests with
a brief description of your project to csm@ncsa.uiuc.edu.
All new users are strongly encouraged to visit NCSA's ANSYS and High-Performance
Computing Environment pages for quick start-up information. Send e-mail
to csm@ncsa.uiuc.edu for all
ANSYS related questions.
NCSA has ANSYS Academic/ Research
license with no restrictions in domain sizes including available Multiphysics capability. There are limited parallel solver
features supported for this type of license like PCG and ICCG solvers on shared memory with 8 CPU-s max.
High level parallel solvers like AMG are not available under this license.
Commercial use of ANSYS is governed by the NCSA Private Sector Program.
Please contact skoric@ncsa.uiuc.edu for more information.
3. How do I get started with MSC/Nastran at NCSA ?
MSC/Nastran is open to all academic users at NCSA. Users with valid allocations
must be placed into the MSC/Nastran access list. Please mail your requestswith a brief description of your project to csm@ncsa.uiuc.edu.
All new users are strongly encouraged to visit NCSA's MSC/Nastran and High-Performance
Computing Environment pages for quick start-up information. Send e-mail to csm@ncsa.uiuc.edu
for all MSC/Nastran related questions.
Commercial use of MSC/Nastran is governed by the NCSA Private Sector Program.
Please contact skoric@ncsa.uiuc.edu for
more information.
4. How do I run Abaqus in batch?
Interactive limits on NCSA machines are 30 minutes and 1Gb. All jobs that exceed these limits
will be stopped, and should be submitted through the batch system. Script
abaqus_batch
has been written to automate the process of submitting Abaqus
jobs to the batch system.
All input files that run serially on a local workstation can be executed in
parallel on the NCSA machines Array without any modifications. The abaqus_batch script will ask the user
to define the number of processors for the Abaqus job. In general, multi-processor
jobs should be limited to 8-32 processors/cores. Choosing the correct number of
processors/cores is a function of the number of degrees of freedom (dof) within
the FEA domain.. For more information on parallel jobs visit the NCSA
Abaqus
page, (Abaqus in parallel section).
5. How do I run ANSYS in batch?
Interactive limits on NCSA machines are 30 minutes and 1GB. All jobs that exceed
these limits will be stopped, and should be submitted through the batch system.
The ansys
script has been written to automate the process of submitting ANSYS jobs
to the batch system.
6. How do I run MSC/Nastran in batch?
Interactive limits on NCSA machines are 30 minutes and 1GB. All jobs that exceed
these limits will be stopped, and should be submitted through the batch system.
A nastran script has been written to automate
the process of submitting MSC/Nastran jobs to the batch system.
All input files that run serially on a local workstation can be executed in
parallel on NCSA machines without any modifications. The nastran
script will ask the user to define the number of processors for the MSC/Nastran
job. In general, multi-processor jobs should be limited to 2-4 processor. Choosing
the correct number of processors is a function of the number of degrees
of freedom (dof) within the FEA domain. For more information on parallel
jobs visit the NCSA MSC/Nastran page.
Nastran requires at least one command line argument. The most common usage
is: nastran input_file_name
For on-line documentation run patran
and its online help.
7. Does NCSA provide pre and post processing software?
In general, most users develop finite element models using software on their
local workstations. Several codes have built-in pre-post processing capabilities. The list of commercial
pre-post processing software is long and diverse. Therefore, the NCSA does not endorse any
particular pre-post code. Users are expected to develop inputs at their local workstations and transfer the
associated input files to NCSA, run solvers in batch and take adventage of large NCSA's computing parallel resources, and bring back output files and process them on their local workstations.
8. Can I export the a Patran finite element mesh elsewhere?
Yes. You can write out the finite element mesh data in the patran 2.5 neutral
format.
Choose File and then Export.... In the new form the format
should be Neutral. Click on the Neutral Options ... and in the new
form the heading Entity Packets lists the entities that can be written to
the neutral format file.
If you are only interested in the finite element mesh data, then highlight
only the following :
- Nodes
- Elements
- Material Properties
- Element Properties.
Ensure that for
Existing Groups the group which contains the finite
element mesh is highlighted. Then click on OK in that form.
In the original form enter a name in the box marked Create New File and
give it an extension out. Example plate.out. Then click on Apply.
9. Can I display Ansys GUI on my local Windows NT/PC?
Yes. However, Exceed 6.2 or 7.1 or newer must
be installed on your NT/PC box. Exceed is a X-server package for Windows. Ansys
GUI has not been certified with other X-server software packages on Windows.
However, it should work with all UNIX workstations, including Linux.
Use the following procedure:
a. Log on cobalt using ssh or kerberos.
b. Start Exceed on your NT/PC box: Start->Programs->Exceed->Exceed
c. Set your display on co-login1 prompt: setenv DISPLAY your_machine_name:0.0
where your_machine_name is the full domain name or
IP address of your PC/NT box
d. type ansys and start interactive GUI session.
If you have problems displaying Ansys GUI with Exceed you need to
adjust you Exceed settings:
1. Start->Programs->Exceed->Xconfig
2 Double-click on the Screen Definition icon.
3. Change the Window Manager to Native (for Exceed 6.2) or Default to Native
(for Exceed 7.1)
4. Double-click the Performance icon.
5. Select Draft Mode, Batch Requests, and Save Unders.
6. Set Maximum Backing Store to Always.
7. Set Default and Minimum Backing Store to When Mapped.
8. Click OK. Exceed will now ask if you want to perfrom a server reset. Click
Yes.
Ask your system administrator to install and adjust Exceed on your PC/NT box.
10. Can I display Abaqus CAE on my Windows PC?
Yes. However, Exceed 6.2 or 7.1 or newer with
Exceed 3D must be installed on your PC. Exceed is a X-server package
for Windows. Exceed 3D extends the X-server with the remote GL library. Remote
GL allows hardware support on the local PC box. Abaqus CAE has not been certified
with other freeware X-servers for Windows like CYGWIN though you can try to download
and use them on your own. Abaqus CAE should be displayed fine on all UNIX workstations, including Linux with remote GL support like Red Hat
Use the following procedure to display CAE on your PC with Exceed:
a. Log on cobalt or abe as usual, with kerberos or ssh.
b. Start Exceed on your PC->Start->Programs->Exceed->Exceed
c. Set your display on prompt: setenv
DISPLAY your_machine_name:0.0
where your_machine_name is the full domain name or
IP address of your PC
d. Run Abaqus CAE with abaqus cae.
Ask your system administrator to install Exceed and Exceed 3D on your PC.
11. Can I run my Abaqus 5.8 input decks with Abaqus 6.5
and 6.6?
Yes. However, Abaqus/Upgrade might be needed to upgrade an input file based
on syntax of Version 5.8 through version 6.1 to the synatx of 6.3, and Abaqus/Free
utility that converts fixed format input files used with previous versions of
Abaqus to 6.6 free format before running with 6.6.
12. ABAQUS ERROR: The following files can not be located
.mdl .stt
This error in .log file is common for restart batch jobs or batch jobs that
require access to non standard abaqus files like external data files or extra fortran/c files for user subroutines.
Make an archive file of all nonstandard files that are missing on batch host
ie:
tar -cvf file2.tar file.mdl file.stt
and then invoke abaqus_batch script:
with additional submit option --datafile2 file2.tar
This option is already included into standard abaqus_batch
script.
13. Is Abaqus/Viewer available ?
No Viewer is a subset of Abaqus/CAE. Please use Abaqus/CAE instead.
14. How do I transfer files from/to my local PC ?
Use the New File Transfer Window icon from SSH menu, or kerberos ftp that is a part of kerberos 5 distribution that you had downloaded to access NCSA machines.
To use kerberos ftp :
Use lcd with "\" notation to change to directories on your
local PC.
Use cd with "/" notation to change to directories on remote
host.
Use open name.ncsa.uiuc.edu to connect to remote host.
Use get file_name to transfer a file from remote host to your
PC.
Use put file_name to transfer a file from a local PC to remote
host.
Use ascii to initiate ascii ftp protocol for ascii files.
Here is an example of an ftp session. Want to transfer config.ps file that
is currently located in C:\myd dir on my local PC to /scratch/users/skoric/csm
dir on the remote machine cu12.ncsa.uiuc.edu.
Start -> Programs -> Kerberos5 -> ftp
(this should bring a black ftp window)
ftp> open cu12.ncsa.uiuc.edu
Connected to cu12.ncsa.uiuc.edu.
220 Cu12 FTP server (Version 5.60) ready.
334 Using authentication type GSSAPI; ADAT must follow
GSSAPI accepted as authentication type
GSSAPI authentication succeeded
Name (cu12.ncsa.uiuc.edu:skoric): <press RETURN here>
232 GSSAPI user skoric@NCSA.EDU is authorized as skoric
202 PASS command superfluous.
Remote system type is UNIX.
Using binary mode to transfer files.
ftp> lcd C:\myd
Local directory now C:\myd
ftp> cd /scratch/users/skoric/csm
250 CWD command successful.
ftp> put config.ps
local: config.ps remote: config.ps
227 Entering Passive Mode (152,153,42,22,138,131)
150 Opening BINARY mode data connection for config.ps
226 Transfer complete.
431062 bytes sent in 0.047 seconds (9e+003 Kbytes/s)
ftp> ls
227 Entering Passive Mode (152,153,42,23,269,42)
150 Opening ASCII mode data connection for /bin/ls.
total 888
-rw-r----- 1 acr 431062 Feb 27 15:26 config.ps
ftp> quit
15. I am getting: "/usr/bin/X11/Xauth: error in locking
..", "over disk quota ..."?
Those messages indicate that you are exceeding disk space in your home directory
that is
1 GB. To check this execute: quota -v
You need to either delete large files in your home directory, ftp them to
Mass
Storage
space, or ftp them to your local machine.
Since most of large simulations are producing large output or temporary files,
it is highly
advisable to run and submit your large jobs from scratch spaces:
/scratch/users/your_login, or (cd scratch-global@ while in your home dir on IBM p690.)
Note that scratch spaces are purged regularly, therefore make sure to
copy all your output files to secure place before they get purged. Most of
CSM batch scripts are automatically transfering all your output files to
your account on Unitree, once your batch job is finished.
16. How do I check the progress of my running batch jobs on cobalt and Abe ?
qstat -u your_login (to get batch id
number of your running job)
cdjob your_batch_id_number (to cd to the working
directory on the batch host)
cd abaqus_batch (for abaqus batch jobs)
or
cd ansys_batch (for ansys batch jobs) or
cd lsdyna_batch (for ls-dyna batch jobs) or
cd nastran_batch (for nastran batch jobs)
Now your are in directory on the batch host where your job is running.
You can check (read only) all files as if you were running interactively and
examine progress of your job.
For example: I want to check the status of my running Abaqus batch job:
qstat -u skoric
the uotput is something like :
23356.co-master skoric standard 23356 test --- --- 4 501760 12:00 R 4:45
(R means my job is running, and the jobid number is 23356)
cdjob 23356
cd abaqus_batch
(now I am in the directory where my Abaqus batch job is running
and can check Abaqus .sta file for progress).
17. My Nastran DMP job exits with SYSTEM FATAL MESSAGE 471 in my .f06 file ?
This is probably due to DMP unsupported INIT DBALL and ASSIGN DBALL statements in your input deck. Please comment them and resubmit your DMP Nastran job.
18. My batch job ran out of wall clock time on cobalt, can I access temporary files ?
If they are not purged from the scratch space on the batch host, please see Q16 to access those files. You should have read access there, and can copy those files to your scratch space, or transfer them manualy to the Unitree. New command saveafterjob always saves your files to Unitree even if your job exits prematurely. This command is implemented in most of CSM batch scripts.
19. How Do I import Pro/E parts into ansys 8.0 on copper?
1. Open you part in Pro/E GUI, and use ANSYS8.0/ANSYSGeom menu from PRO/E GUI.
This will produce name.anf file in your working dir, where name is the name of your proe part.
2. mv name.anf name.prt.anf (ie. rename your file).
3. Start ansys GUI, from import/proE choose name.prt.anf
and a small window with ansysconproe message will pop up, press proceed in it.
20. My interactive job under GUI exceeds interactive limits on cobalt, what should I do ?
Running GUI remotely is slow and prone to large
network latencies. If you still want to do dit, you must perform procedure from Q9 or Q10 or Q25 , ie. to have xclient running on your local machine and setenv DISPLAY on co-login.ncsa.uiuc.edu to your local machine or to have X11 tunneling enabeled with your ssh. Then invoke single cpu interactive job qsub -I . You will wait in queue as for any other batch job untill your job starts running, check it with qstat -u your_login . Once your job starts running, you will be dispatched to a compute host of cobalt and you can invoke gui the same way you do it on interactive machine co-login1. You should be able to have access to the batch host and run GUI according to the time and memory limits you specified with qsub -I . The waiting time in queue depends on cpu resources you request, the more you request, the longer you will wait. You must be ready to use it since you will be charged for the time when interactive job starts, whether or not you are using it. Therefore, for larger access use this facility only out of prime time hours to decrease waiting time in queue.
21. My Abaqus job crashes with: executable standard.x aborted with system error "Illegal Memory References (signal11) ."
This used to be a known bug with Abaqus/Standard 6.4 that sometimes occured for contact or plasticity problems. It should be resolved with version 6.5 and up. Please report it, if you experience it on NCSA machines. Other common occurence of this problem is for Abaqus runs with user defined subroutines UMAT and VUMAT, and usually something is wrong with your UMAT or VUMAT coding (like for example the divison by zero occuring for some loading case) causing this error.
22. Can I run Abaqus solver from CAE GUI on NCSA machines ?
NO, Please Save your .inp file in CAE and run it with Abaqus command line interactively, or in batch with abaqus_batch .
23. I ran my Ansys job in batch. How do I postprocess results ?
If you have a correct Ansys syntax (APDL language) in your input file that performs preprocessing and Ansys solution and than save everything properly, you will have a result file (.rth or .rst file) copied back to the submit directory. Then you just have to display Ansys GUI, Main Menu -> General Postproc. -> Data & File Opts brings up the Data and File Options dialog box. The results file is selected by clicking on the button with theree dots. After selecting the correct file, click on OK and perform the postprocessing. It the solution involves multiple subsets and load steps. there are many result sets and the user should select the correct (desired) one
24. Fonts are too small and not readible in Ansys GUI ?
Crete a .Xdefaults file in your home directory with the following lines :
*EUIDL*font: Times 16 normal
*EUIDL*Menu*font: Courier 16 bold
*EUIDL*Menubutton*font: Courier 14 bold
*EUIDL*MainMenu*font: Times 14 {normal italic}
Save the .Xdefaults file and logout and login to have the new defaults take effect.
25. How do I run GUIs on Abe ?
Running GUI remotely is slow and prone to large
network latencies. If you still want to do it please check with your network support that your personal
or your network firewall is not blocking Xwindows traffic.
In your ssh gui under Edit Settings make sure that:
- "authentication" is set to accept ssh2 connections.
- "tunneling" has X11 tunneling set.
- save the new ssh settings.
1. Start Exceed (Exceed 3D for Abaqus/CAE) on your
PC, and then log on abe with x11 tunneled.
2. Open an xterm, ie type on login node:
xterm &
A new window with an xterm will pop.
3. In that new window, submit an interactive
job, ie type:
qsub -I -V -l walltime=01:00:00,nodes=1:ppn=8
-walltime hh:mm:ss specifies max wall clock time access to a
Abe's compute node.
Eventually you will be dispatched to a compute node in this window.
Name of the compute node will be given at the prompt.
4. Now, in the original login access window type :
ssh -X -Y name_of_compute_node_where_your_job_is_dispatched
and run X Applications from there like :
abaqus cae & or
ansys &
26. I am getting qsub: illegal -N value on cobalt and my job is not submitted ?
The name of your PBS batch job is longer than 14 characters or starts with a number. This name is your input file name by default when submitting with abaqus_batch script. Please simplify it to comply with this PBS naming rule.
