ARCHIVED: Rockhopper at Indiana University
On this page:
- System overview
- System information
- System access
- Available software
- Computing environment
- Transferring your files to Rockhopper
- Application development
- Running your applications
- Reference
- Policies
- Support
System overview
Rockhopper (
rockhopper.uits.iu.edu
) is Penguin
Computing's Penguin-On-Demand (POD) supercomputing cloud appliance
hosted by Indiana University. The Rockhopper POD is a collaborative
effort between Penguin Computing, IU, the University of Virginia, the
University of California Berkeley, and the University of Michigan to
provide supercomputing cloud services in a secure US
facility. Researchers at US institutions of higher education and
Federally Funded Research and Development Centers (FFRDCs) can
purchase computing time from Penguin Computing, and receive access via
high-speed national research networks operated by IU.
Rockhopper consists of 11 Penguin Computing Altus 1804 servers, each containing four AMD Opteron 6172 12-core processors and 128 GB of RAM. The total RAM in the system is 1.5 TB. Each server chassis has a QDR (40 Gbps) InfiniBand interconnect to the cluster's switch fabric, which is then connected via four trunked 10 Gbps Ethernet links to IU's network infrastructure. For hardware configuration details, see this document's System information section.
The Rockhopper nodes run CentOS 5. Job management and scheduling are provided by the Sun Grid Engine (SGE) resource manager. The Modules system is used to simplify application and environment configuration. Users may log into the cluster via SSH, using their Penguin POD user IDs.
System information
System configuration | Aggregate information | Per-node information |
---|---|---|
Machine type | High-performance, on-demand computing Penguin Computing Altus 1804 MPP cluster |
4 x 2.1 GHz 12-core AMD Opteron 6172 processors |
Operating system | CentOS 5 | |
Memory model | Distributed | |
Processor cores | 528 | 48 |
CPUs | 2.1 GHz 12-core AMD Opteron 6172 processors | |
Nodes | 11 compute nodes; 2 login nodes | |
RAM | 1.4 TB 1,333 MHz DDR3 ECC memory | 128 GB 1,333 MHz DDR3 ECC memory |
Computational systems details | Total | Per node |
Processing capability | Rmax = 5242 gigaFLOPS | Rmax = 403 gigaFLOPS |
Benchmark data | HPL gigaFLOPS 2216.53 (8 nodes) | HPL gigaFLOPS 288.67 |
Power usage | To be determined | 0.000230 teraFLOPS per watt |
Disk space | 67 TB (local) | 6 TB (local) |
Login nodes | Rmax = 806 gigaFLOPS | Rmax = 403 gigaFLOPS |
Homogeneous compute nodes | Rmax = 4435 gigaFLOPS | Rmax = 403 gigaFLOPS |
Heterogeneous compute nodes | Rmax = 4435 gigaFLOPS | |
Storage information | ||
File systems | Home directories are on a local Lustre file system with no quotas. | |
Disk space | 67 TB (local) | |
Total scratch space | Rockhopper does not include a separate scratch file system. Disk usage is a billable item. | |
Data Capacitor II | Note: Shared
scratch space on Data Capacitor II is available only to IU
students, faculty, and staff.
Shared scratch space is hosted on the Data Capacitor II (DC2) file
system. The DC2 scratch directory is a temporary workspace. Scratch
space is not allocated, and its total capacity fluctuates based on
project space requirements. The DC2 file system is mounted on IU
research systems as |
|
Scholarly Data Archive (SDA) | Note: Archival storage space on IU's HPSS-based Scholarly Data Archive (SDA) is available only to IU students, faculty, and staff | |
Backup and purge policies | Home directories are not backed up. | |
Quotas | None | |
Networking Information | ||
QDR (40 Gbps) InfiniBand | Switch: Mellanox IS5030 HCA: Mellanox ConnectX 2, 1-port QSFP, QDR 1 GBps Ethernet |
System access
Requesting an account
Researchers at US institutions of higher education (with
.edu
domain names) or Federally Funded Research and
Development Centers (FFRDCs) can
purchase computing time from Penguin Computing, and
then receive access to the on-demand HPC cloud service hosted on
Rockhopper at IU. Prospective users can request accounts by filling
out and submitting Penguin Computing's
account request form. To pay for your account, you need to enter
your credit card information when completing the account request
form. To request an alternate financial arrangement, email Penguin
Computing directly.
Methods of access
Users may log into the Rockhopper POD via SSH, using their Penguin POD user IDs. Users create SSH keys and obtain instructions on how to use them as part of the account creation process.
SSH2 clients may be used to connect to
rockhopper.uits.iu.edu
, which resolves to one of
Rockhopper's two login nodes:
login0.rockhopper.uits.iu.edu login1.rockhopper.uits.iu.edu
File transfer is supported via SCP or SFTP.
Note: Public key authentication is the only permitted authentication mechanism available on the Rockhopper cluster:
- Rockhopper does not use local passwords/passphrases for user authentication.
- Rockhopper does not use IU ADS for authentication. Rockhopper is not an IU resource; you cannot use your IU Network ID to access the POD system.
Available software
Software packages installed on Rockhopper are made available to users via Modules, an environment management system that lets you easily and dynamically add software packages to your user environment. For a list of software modules available on Rockhopper, see Rockhopper Applications.
For more about using Modules on Rockhopper, see Modules in the Computing environment section of this document.
Computing environment
Unix shell
Rockhopper supports the Bourne-again (bash
) and Tenex
C (tcsh
) shells. New user accounts are assigned the
bash
shell by default. For more on bash
, see
the Bash
Reference Manual.
To change your shell to tcsh
on Rockhopper, email Penguin
Computing directly.
Environment variables
Use the following commands, depending on your shell, to display and change the values of environment variables:
Shell | Display value | Change value |
---|---|---|
bash
|
echo $VARNAME
|
export VARNAME=VALUE
|
tcsh
|
echo $VARNAME
|
setenv VARNAME VALUE
|
Startup scripts
- Bash: On login,
bash
reads and executes commands from the following directories by default (and in this order):/etc/profile ~/.bash_profile ~/.bashrc
Note: The
~
(tilde) represents your home directory (e.g.,~/.bash_profile
is the.bash_profile
file in your home directory).On logout,
bash
reads and executes~/.bash_logout
. For more onbash
startup files, see the Bash Startup Files section of the Bash Reference Manual. - Tcsh: On login,
tcsh
reads and executes commands from the following directories by default (and in this order):/etc/csh.cshrc /etc/csh.login ~/.tcshrc (if it exists, otherwise ~/.cshrc) ~/.history ~/.login ~/.cshdirs
Note: In practice, on Rockhopper, only the first two files exist. You may create the others, and add commands and variables to them as you see fit.
Modules
Rockhopper uses the Modules environment package to provide a convenient method for dynamically modifying your environment and specifying which versions of software you use.
Some common Modules commands include:
Command | Action |
---|---|
module avail
|
List all software packages available on the system. |
module avail package
|
List all versions of package available on the system;
for example:module avail openmpi |
module list
|
List all packages currently added to your user environment. |
module load package
|
Add the default version of the package to your user
environment; for example:module load intel |
module load package/version
|
Add the specified version of the package
to your user environment; for example:module load intel/11.1 |
module unload package
|
Remove the specified package from your user
environment.
|
module swap package_A package_B
|
Swap the loaded package (package_A ) with another
package (package_B ). This is synonymous with:module switch package_A package_B |
module show package
|
Show the changes loading the specified package makes
to your user environment (e.g., environment variables set, library
paths added). This is synonymous with:module display package |
For more, see Use modules to manage your software environment on IU research supercomputers.
Transferring your files to Rockhopper
Rockhopper supports SCP and SFTP for transferring files. SCP is a command line utility included with OpenSSH. Basic use is:
scp -i ~/.ssh/key-you-downloaded-from-pod [[user@]host1:]file1 [[user@]host2:]file2
For example, to copy foo.txt
from the current
directory on your computer to your home directory on Rockhopper, use
(replacing username
with your Rockhopper username):
scp -i ~/.ssh/key-you-downloaded-from-pod foo.txt username@rockhopper.uits.iu.edu:foo.txt
You may specify absolute paths, or paths relative to your home directory:
scp -i ~/.ssh/key-you-downloaded-from-pod foo.txt username@rockhopper.uits.iu.edu:/some/path/for/data/foo.txt
You also may leave the destination filename unspecified, in which case it will become the same as the source filename. For more, see Use SCP to securely transfer files between two Unix computers
SFTP provides file access, transfer, and management, and offers client functionality similar to FTP. For example, from a computer with a command line SFTP client (e.g., a Linux or Mac OS X workstation), you could transfer files as follows:
$ sftp -i ~/.ssh/key-you-downloaded-from-pod username@rockhopper.uits.iu.edu: Connected to rockhopper.uits.iu.edu. Changing to: /home/username/ sftp> ls -l -rw------- 1 username group 113 May 19 2011 loadit.pbs.e897 -rw------- 1 username group 695 May 19 2011 loadit.pbs.o897 -rw-r--r-- 1 username group 693 May 19 2011 local_limits sftp> put foo.txt Uploading foo.txt to /home/username/foo.txt foo.txt 100% 95MB 8.7MB/s 00:11 sftp> exit
You can also ship detachable hard drives to Penguin Computing. To make such an arrangement, email Penguin Computing directly.
Graphical SFTP clients are available for many systems. For more, see Use SFTP to transfer files
Application development
Programming models
Rockhopper is designed to support codes that have reasonably large shared memory and/or distributed memory parallelism.
Compiling
The GNU Compiler Collection (GCC), Intel Compiler Suite, and
Portland Group (PGI) compilers for C, C++, and Fortran codes are
installed on the Rockhopper POD. Open MPI wrappers for these compilers
are available for MPI programs. Use the -O3
switch for
optimization of serial and parallel codes. The Intel Math Kernel
Library (MKL)
and AMD Core Math Library (ACML)
are available. For debugging, the GNU Project Debugger (GDB) and Intel Debugger
(IDB)
are available.
For more about compiling programs on the Rockhopper POD, see Compiling on POD IU on Penguin Computing's POD IU wiki page.
Running your applications
Rockhopper has one general-purpose queue (all.q) with "First In First Out" (FIFO) scheduling and no maximum walltime limit.
Rockhopper uses the Sun Grid Engine (SGE) for job management and
scheduling. If you're familiar with TORQUE (or OpenPBS),
you'll most likely be comfortable working with SGE. Many job
submission (qsub
) parameters are identical in TORQUE and
SGE; the only difference is SGE replaces #PBS
with
#$
. For complete user information, see the Grid
Engine manual pages.
Penguin Computing also provides the PODTools package for submitting and managing jobs from your personal workstation without connecting to the Rockhopper login node. The package includes POD Shell for remote job submission, data staging tool, and file management, and POD Report for querying your core-hour and storage usage.
Submitting jobs
To submit a job to run on the Rockhopper POD, use the
qsub
command. If the command exits successfully, it will
return a job ID, for example:
[jdoe@login1 ~]$ qsub job.script Your job 2651 ("job.script") has been submitted [jdoe@login1 ~]$
If you need attribute values different from the defaults, but less
than the maximum allowed, specify these either in the job script using
SGE directives, or on the command line with the -l
switch. For example, to submit a job requiring 10 hours of walltime,
use:
qsub -l time=10:00:00 job.script
Note: Command-line arguments override directives
in the job script, and you may specify many attributes on the command
line, either as comma-separated options following the -l
switch, or each with its own -l
switch. The following
commands are equivalent:
qsub -l time=10:00:00, mem=1G job.script qsub -l time=10:00:00 -l mem=1G job.script
Useful qsub
switches include:
Switch | Action |
---|---|
-q <queue_name>
|
Specify a user-selectable queue. |
-r
|
Make job re-runnable. |
-a <date_time>
|
Execute the job only after a specified date and time. |
-V
|
Export environment variables in your current environment to the job. |
For a complete list of qsub
parameters, see this
qsub
manual
page.
Monitoring jobs
Use qstat
for monitoring the status of a queued or
running job. Switches include:
Switch | Action |
---|---|
-u <user_list>
|
Display jobs for users in the user list. |
-sa
|
Display all jobs. |
-sr
|
Display running jobs. |
-r
|
Display the full listing of jobs (excessive detail). |
-n
|
Display nodes allocated to jobs. |
For a complete list of qstat
parameters, see this
qstat
manual
page.
Deleting jobs
Use qdel
to delete queued or running jobs.
Occasionally, a node will become unresponsive and unable to respond to
SGE's requests to kill a job. In such cases, try qdel
-f
.
For a complete list of qdel
parameters, see this
qdel
manual
page.
Reference
User and support information is available on Penguin Computing's POD IU wiki page.
Policies
Accounts
Researchers at US institutions of higher education (with
.edu
domain names) or Federally Funded Research and
Development Centers (FFRDCs) can
purchase computing time from Penguin Computing, and
then receive access to the on-demand HPC cloud service hosted on
Rockhopper at IU. Prospective users can request accounts by filling
out and submitting Penguin Computing's
account request form. To pay for your account, you need to enter
your credit card information when completing the account request
form. To request an alternate financial arrangement, email Penguin
Computing directly.
Acceptable use
The Rockhopper cluster is a Penguin Computing resource. For information about your responsibilities as a user of this resource, see POD Terms of Use.
Home directories
Home directories reside on a Lustre file system, with no quotas or backups.
Note: Rockhopper does not provide a separate scratch file system. Disk usage is a billable item; to make financial arrangements, email Penguin Computing directly.
Computational resources (queues)
Rockhopper has only one queue, and all jobs submitted will execute in the default queue. The only restriction is that individual jobs are limited to 128 cores.
Mail usage
Rockhopper does not provide a production mail service; however, SGE
communicates via email. Mail forwarding is not configured during
account creation; you should consider establishing a mail forwarding
.forward
file; see ARCHIVED: How do I forward my mail from a Unix account?
Scheduled downtime
Rockhopper does not have a regularly scheduled maintenance window. Information about pending outages is sent via email to account holders.
Support
For user and support information, see Penguin Computing's POD IU wiki page, or email Penguin Computing directly.
This is document bbjb in the Knowledge Base.
Last modified on 2021-05-04 15:44:36.