From CCS Cluster Wiki
CCS Linux Clusters
If you have questions that are not addressed within the wiki please contact email@example.com.
There are two clusters available for job submission at the CCS. Sphynx is our newest cluster with 576 cores, homogeneous hardware, and QDR Infiniband. Ares, our older cluster, has 252 cores spread out across two, four, eight, and sixteen core nodes.
The two clusters' technical specifications are listed below:
- Cores: 576 total cores (192 2.4GHz AMD Opteron 6136, 192 2.6GHz AMD Opteron 6212, 192 2.8GHz AMD Opteron 6220)
- FLOPS: 5.9 TFLOPS
- Network bandwidth: 40Gb/sec Infiniband
- Memory: 2.25TB (4GB/core)
- Disk Array: 20TB via 40Gb/s Infiniband
- Cores: 180 2.0-2.4 GHz AMD Opteron
- Network bandwidth: 10Gb/sec Infiniband
- Memory: 504GB (2GB/core)
- Disk Array: 20TB via 1Gb/s Ethernet
Obtaining an Account
To obtain an account with the CCS you must first fill out a resource request form and submit it for review here.
Changing Your Password
Log into Sphynx and run the passwd command to change your password. Replace remoteuser with your CCS cluster username.
user@localhost> ssh firstname.lastname@example.org
Last login: Tue Aug 14 13:50:50 2012 from 127.0.0.1
Changing password for user remoteuser.
Enter login(LDAP) password:
Retype new password:
LDAP password information changed for remoteuser
passwd: all authentication tokens updated successfully.
General Linux Usage
Batch Queuing System
Users' home directories are located in either /u00 or /u01. Home directories have a very small quota and should only be used for storing source code and notes.
Jobs run on the cluster should be configured to write output data within your /scratch03/user directory.
Compute nodes in the cluster do not have permission to write to your home directory.
The CCS does not keep backups of data in scratch directories.
Much of the software available to you on ares requires many environment variables to be set. To simplify this a command line utility named module is available.
- PETSc - Portable Extensible Toolkit for Scientific Computation
- Eigen3 - C++ template library for linear algebra: matrices, vectors, numerical solvers, and related algorithms
Advanced Cluster Topics
Consult the User's Guide for information on using the wiki software.