Scientiﬁc computing resources at FHCRC

Jerry Davison and Martin Morgan

06 March 2014

For updated content, please see the Computing section of the Biomedical Data Science Wiki

____________________________________________________________________________________________

1 Scientiﬁc Computing’s resources run the Linux operating system
1.1 Access
1.2 Logging in to a rhino
1.3 Starting and using R from the shell

_______________________________________________________________________________________________

1 Scientiﬁc Computing’s resources run the Linux operating system

Many high-performance computers here and elsewhere run the Linux operating system. Interaction with it to access and manipulate ﬁles is through text commands. There are plenty of web sites oﬀering help, here are two, and a publisher:

Linux ABC’s
Linux for newbies
O’Reilly Linux in a Nutshell - get the 1st or 2nd edition, later ones are badly bloated.

The Scientiﬁc Computing group (http://centernet.fhcrc.org/CN/depts/it/scientific_computing/) at the Hutch provides high-performance computing access to FHCRC researchers, without charge. These computers enable analyses diﬃcult or impossible to perform on a typical laptop; all run the same version of the Ubuntu Linux operating system. There are four rhino machines, with these capabilities:

multicore – 16 cores (cpu’s)
large memory – 378 GB main memory
large secondary storage – 4 TB /home hard drive shared among users, plus additional large /shared volumes for data storage

1.1 Access

There are several ways to access a rhino:

MAC OS X and Linux: open a terminal and log in with the secure shell command

ssh -X myHutchID@rhino

and then enter your password. You’ll be connected to one of the rhino’s. The -X option enable X11 forwarding – so that plots can be presented on your laptop .

MS Windows users: use PuTTY – get putty.exe from http://www.chiark.greenend.org.uk/~sgtatham/putty/download.html to open a rhino terminal, and install Xming (http://sourceforge.net/projects/xming/) to process R plot commands – so plots can be presented on your laptop.

The ﬁrst time you run PuTTY you’ll need to navigate through the Category menu on the left to Connection, SSH, and X11; click on the Enable X11 forwarding box. Return to Session and Save the Default Settings. In the Host Name box enter rhino04 and click on Open. Log in with your Hutch ID and password. To enable plotting you need to start Xming.

1.2 Logging in to a rhino

When logging in to one of the rhino’s you can specify a particular rhino (rhino01, rhino02, rhino03, or rhino04). This text is printed to your screen when you log in:

Welcome to Ubuntu 12.04.3 LTS (GNU/Linux 3.5.0-43-generic x86_64)
############################################################
                    IMPORTANT NOTES:
         (************ UPDATED ************)
* This computer is intended for interactive scientific computing,
  development and for compute jobs that require > 20GB memory.

* If you are running a long & CPU intensive job (> 1h) and your
  job requires less than 20GB of memory you can start an inter-
  active session to Gizmo by using the 'grab*' commands to get
  immediate access to a compute server with 1-12 cores.

  However in most cases it will be more effective if you use the
  sbatch commmand.

  * Shared folders can be found in:
        /shared/silo_researcher
        /shared/labs

* To use this system most effectively, please read this summary:
    http://scicomp.fhcrc.org/Gizmo%20Cluster%20Quickstart.aspx

############################################################

Do have a look at the documentation at http://scicomp.fhcrc.org/Gizmo\%20Cluster\%20Quickstart.aspx and login for an introduction to using the scientiﬁc computing resources, especially if you plan to run long analyses. If you have questions about using them, send email to scicomp@fhcrc.org.

1.3 Starting and using R from the shell

To start an R session, at the shell prompt enter “R”:

myShellPrompt$ R

> plot(1:10, type=’h’)
> q()

To use R studio instead, enter:

myShellPrompt$ rstudio