Getting started with XSEDE

Getting started with XSEDE

Most of our computation will be done on comet, which sits within SDSC but is part of a larger supercomputing platform called XSEDE. This document gives quick info on how to get started.

Getting an XSEDE account

Go to the XSEDE portal and create an account.

Log into comet

ssh [email protected]

Show which accounts you have allocations for

show_accounts

This will show something like:

ID name      project      used     available    used_by_proj
------------------------------------------------------------`
mgymrek      ddp268       0        500000       0       
mgymrek      sds154       0        1000         70302

You can get more details about a project by doing:

proj_details ddp268

   userid        spent     can spend          real name        
------------   ---------   ---------  -------------------------
  anz023          0       500000  An Zheng
  agoren          0       500000  Alon Goren
 mgymrek          0       500000  Melissa Ann Gymrek

Filesystem

Right now there are three places you can put things:

1. In the scratch directory, which has no hard quota but is for temporary files

   /oasis/scratch/comet/$USER/temp_project

2. In the project directory, which has a hard 0.5T limit. Note "ddp268" is our project ID, which will be different for different allocations.

   /oasis/projects/nsf/ddp268

3. You can store small things like source code in your home directory:

   /home/$USER

Submitting jobs

See the XSEDE documentation