Getting an Account
When you’re ready to start computing, you’ll open an account. There is no fee for getting an account or using the publicly available resources of the NNIN/C@UM. When you get an account from the NNIN/C@UM, you’ll get accounts our public machine, currently restricted to MNC2 (Michigan Nano Computational Cluster). You can test your code on our system on which you have an account.
Academic Users Inside the University of Michigan
UM applicants are required to have a U of M faculty sponsor. UM students, from any college, contact us if you do not have a sponsor and we may be able to locate one for you. U of M students please use the following form to request a local account:
Users from Outside the University of Michigan
If you are not working with a University of Michigan researcher, then you need to get a U-M Sponsored Uniqname. Please use the following form to request a Uniqname and NNIN/C account:
For further information on NNIN/C@UM accounts, e-mail: bshiari@umich.edu or phone (734) 615 4804.
Training
The NNIN/C@UM offers training several times a year on a variety of topics. Training is announced on the website and in email to users with valid accounts. If there is a group of 5 or more the NNIN/C@UM will offer an additional training outside the normal schedule, contact the NNIN/C@UM to schedule. All training materials for reference, are available for download.
Using Your Account
You have to learn how to transfer your data, what a PBS script looks like, and how to submit your job. Use a recommended NNIN/C PBS.
A quick reference guide on different commands (unix, pbs, etc) that you’ll want to know to use of the NNIN/C@UM system can be found here (link to Running Jobs; Submitting Jobs via PBS).
In general, the steps to using the NNIN/C@UM cluster/resources:
1. Transfer your data to the cluster using scp/sftp.
2. Login into the cluster (mnc2.eecs.umich.edu). If you’re logging in from off-campus, you’ll want to use ssh.
3. Write a PBS script that does what you want to accomplish. PBS scripts are shell scripts with some directives for the resource manager (job length, number of processors, amount of memory, etc.) Examples of PBS scripts can be found here.
4. Submit your job with qsub: qsub name_of_your_pbs_script
5. Cluster scheduler will find appropriate resources to run your job.
6. If you set the email flag in your PBS script, you’ll receive an email when your job starts and/or ends.
