HFSS
Last Modified: 08/10 9:55
Description
From Wikipedia: HFSS is a commercial finite element method solver for electromagnetic structures from Ansoft Corporation.
Version
- 11.1
Authorized Users
- Students doing research
- Faculty
Platforms
- Cluster
Instructions for First-Time Use
You will need to use the modules program, so if you're not familiar with it, check out the wiki link.
- Before you login to circe, you need to make sure you have X11 Forwarding enabled. If you're logging in from a UNIX/Linux system, you will need to do the following:
ssh -Y user@circe.rc.usf.edu
If you're logging in from a Windows machine, you need to Enable X11 Forwarding in whatever ssh client you're using. Make sure you've followed our XWin32 Guide to configure your connection to Circe.
- Make sure you're logged into either login0 or login1. (So you know, whenever you ssh to circe, it actually logs you into one of those machines)
- Run the following:
module add apps/ansoft/hfss-11
- Next, run this:
module initadd apps/ansoft/hfss-11
That will run the HFSS module every time you login to any machine on our cluster.
- Once that's done, run an interactive job by typing the following:
qlogin
That will give you a shell directly on a compute node. You will have to run this command (with additional options) whenever you run HFSS in the future. For now, we will use the default 10-minute session that qlogin gives us to configure HFSS.
- Next, run HFSS itself by typing hfss.
- Two dialog boxes will pop up. Just click OK on them. The third box will ask you what directories you want to store your projects and temporary files in. You shouldn't have to change the path that the temporary files will be stored at, but you might want to change the path where the project files will be stored. Fill that out, hit OK, and HFSS will run.
You will only have to do the previous set of steps once. For regular usage, check out the section below.
Interactive Execution on CIRCE
This method assumes that you wish to use the HFSS design graphical interface while running your analysis. For shorter analysis (<= 1 hour), this is an ideal method although one must consider that there are only a handful of licenses available to use the HFSS desktop application. Also consider that the responsiveness of the HFSS user interface while running over X11 forwarding is less than ideal.
- As above, make sure X11 Forwarding is enabled before logging into circe
- Now you will want to start an interactive job, but unlike the first-time instructions, you do NOT want to just run qlogin with no options. This is because the default session time limit is 10 minutes, and you won't get to choose which queue you run your job in. Please refer to this wiki to learn how to run interactive jobs properly.
- Once you login to the node, simply run hfss.
Batch Execution on CIRCE
This is the preferred method for running HFSS on CIRCE, avoiding the need to do X11 forwarding. You must ensure that you've gone through the First-time instructions above. Batch execution may seem daunting to the uninitiated, but it can be an invaluable tool for running models for the following reasons:
- Batch jobs are scheduled when enough licenses are available. This frees you from having to worry about other users using the software and if a license will be available for you to run the solver
- Facilitates easier remote execution. Rather than having to start an X11 session over your internet connection at home, a simple command-line session is all you will need to get started. This saves you from the frustration of slow window refreshes, etc.
- You may not always be able to get resources with qlogin, especially if the system is busy. This means that you will need to return periodically to see if there are available resources to use. With batch mode, this problem is solved since the scheduler handles stopping and starting your job for you. When execution begins and ends, you'll receive an e-mail notifying you, every step of the way.
To use batch mode, you will need to set up your Project, preferably on your workstation or desktop, up to the point where you would run one of your analyses. When you get to this point, save your project. It will be stored in a <name_of_model>.hfss file. You should copy this file with your SSH client or WinSCP to your CIRCE account. For the sake of organization, you should probably place the file in ~/Ansoft, the Ansoft directory in your /home directory.
- Change your directory to ~/Ansoft
- Choose the project for which you wish to run analyses. In this example, we'll use antenna1.hfss.
- Run the job submission client, hfss-batch:
[user@host ~]$ cd ~/Ansoft [user@host Ansoft]$ hfss-batch antenna1.hfss Creating submission request for "antenna1.hfss" 1. Specify the design you wish to analyze (Leave blank for all/any designs): 2. Specify the mode you wish to use for your analysis (Nominal,Optimetrics or leave blank for both): 3. Specify the analysis setup wish to use (leave blank for any/all setups): Setup1 4. Specify the estimated runtime in [HH]HH:MM:SS for this job: 48:00:00 5. Specify the number of processors you wish to use for this analysis [1-16]: 16 6. Specify the amount of memory in Megabytes (M) or Gigabytes (G) that you will need: 28G 7. If you wish to be notified of job start/stop via e-mail, type your address here: user@email-domain.com Your job 1000 ("antenna1.hfss") has been submitted Congratulations! You have submitted your HFSS batch job for "antenna1.hfss"! To check the running status of your job, you can use the "qstat" command. - Check on the status of your job by running qstat:
[user@host Ansoft]$ qstat job-ID prior name user state submit/start at queue slots ja-task-ID ----------------------------------------------------------------------------------------------------------------- 1000 0.00000 anntenna1. brs qw 07/27/2009 11:43:05 16
Getting Help
Please report any issue to the USF IT Helpdesk by sending an e-mail to mailto:help@usf.edu
