New Group Member Introduction

Introduction

This page provides a brief introduction to the computational infrastructure of the Neuroimaging and Neuroinformatics group (NIG). It consists mainly of links out to various bits of reference material.

The NIG's scientific computing team comprises Neil Killeen (leader, applications, informatics), Wilson Liu (applications, informatics) and Sina Sadeghi (IT and data operations). The NIG operates and supports its own IT infrastructure. Other IT groups which we liaise with includes those of the Florey Neuroscience Institutes, Melbourne University Information Services

The team provides application and IT support for the NIG's facilities location in three places:

• The Neuroimaging LAB in the Allan Gilbert Building, Level 3. This is where the majority of the group is located.
• The basement of the Howard Florey Institute which houses the small animal MR imaging Facility
• The Department of Electrical Engineering which houses the Altix (soma) Supercomputer (IT support from Engineering IT); the group's main computational platform.

Accounts

When you first arrive, accounts for you should largely be setup already. Sina will discuss these with you.

There are a number of accounts:

• General NIG computer account
• soma (super computer, see below)
• Trouble Ticket (for suport requests; see below)
• Data Repository (see below)

Documentation and Getting Support

• The NIG's external web page is here
• The NIG's Intranet is here
• The home page for Scientific Computing Documentation is here
• Please use the web-based Support Ticket system to get help. However, we are always personally available for urgent problems.
• A list of IT related guides (such as how to access the wireless, print from your laptop, etc) can be located here.

Computers

Computers come in a few categories:

• Desktops; you will be provided with a Mac, Linux or Windows desktop depending upon your needs
• File servers; soma is the primary computer for storing data. However, there is a small amount of storage available on other computers (discuss with Sina if required)
• Compute servers; soma is the primary compute server. There are no other mid-range compute hosts apart from your desktop

Applications

• Applications are provided as uniformly as possible across the various Unix computers (including soma)
• For each Unix application that you want to use, a module file is used to provide access. Please read about this here

Data Repository

• We operate a repository for the RAW data that you acquire from MR scanners. We provide a web interface and the system is built on top of a product called Mediaflux
• You can read about the repository here

Accessing soma

• soma is a 64-processor SGI Altix shared-memory computer with a 62TB data store. The data store combines disk and tape into one large virtual file system managed by software called DMF
• Top-level documentation for soma is here
• Access to soma is by ssh ; see here for details
• You can access soma's data store remotely (e.g. from your desktop) as described here but you should do this carefully
• You must read about how DMF works here. It is largely but not always transparent. You should know:
  • How to recall files in bulk from tape
  • How to see a files state (disk/tape/both)
  • How to migrate files to tape
• Jobs are submitted to a queuing system and you must read about this here and you should know:
  • How to submit your jobs to the queue
  • How to view the queue status
  • How to delete jobs from the queue

-- NeilKilleen - 20 Oct 2009