Principal  Investigator (PI)

Startup allocations  are targeted at new users or users who have very modest computational  requirements (less than 10,000 hours). To apply, an eligible Principal  Investigator (PI) must submit an application available on any of  the partner web pages listed below. A full proposal is not normally  required. There is only one startup given to each PI. Requests for startup  awards can be submitted at any time and take about one month to process.

 People entitled to use AHPCC  and Alliance systems: UNM faculty, staff, and students along with their  authorized collaborators; university and government researchers who  submit an appropriate project proposal.

Persons employed by Federally funded Research and Development Centers  (FFRDCs-e.g., NRL, DOE, NASA) are not normally supported. However, a  scientist, engineer or educator who has a joint appointment with a university  and a Federal agency (such as a Veterans Administration Hospital, or  with a university and an FFRDC) may submit proposals through the university  and may receive support if he/she is a bona fide faculty member of the  university, although part of his/her salary may be provided by the Federal  agency. Under unusual circumstances, other Federal agencies and FFRDCs  may submit proposals directly to NSF. Preliminary inquiry should be  made to the appropriate program before preparing a proposal for submission.  You can read more details in the NSF Grant Proposal Guide.

 

An acknowledgement of UNM-Alliance support must appear in publications  of any material, whether copyrighted or not, based on or developed with  resources of the University of New Mexico and the Alliance.

Suggested Wording
"We would like to thank the University of New Mexico for access  to their Albuquerque High Performance Computing Center. This work utilized  the UNM-Alliance Linux Superclusters."

Submitting Copies
Copies of technical reports and publications using AHPCC resources should be sent to AHPCC with the grant number listed with  your login information.

 

Acceptable Use
By accepting an UNM-Alliance Supercluster account, I agree that:

I will be responsible for all use of this computer account. I will use the computer account for High Performance Computing purposes  and not for commercial purposes. I will not use the computer account to engage in any form of illegal  software copying or other copyright infringement. I will not attempt to access accounts, files or information belonging  to other users without their knowledge and consent. I understand the monitoring of user activity on the Albuquerque  High Performance Computing Center machines may occur.

FAILURE TO COMPLY WITH THESE RULES WILL RESULT IN REMOVAL OF THIS  ACCOUNT, AND POSSIBLE DISCIPLINARY OR LEGAL ACTION.

Security
Users are provided an initial Default Password in their new user  packet. You should change your password the first time you login and  periodically thereafter. If it is necessary for you to request a  password change from us (for example, when you have forgotten your current  password), your password will be set to this default. It  is important that you keep this form in a secure place for your reference.  If you do not have this form available, you will be required to wait  until a duplicate can be mailed to you via U.S. mail. AHPCC does NOT  provide passwords by phone, fax, or electronic mail.

 

The following guidelines for the appropriate use of AHPCC computing  facilities are to be followed by all users. These guidelines are meant  to ensure the security of user accounts as well as equitable access  to our computing facilities by all users.

Acceptable Use
 By accepting an UNM-Alliance Linux Supercluster account, I agree  that:

I will be responsible for all use of this computer account. I will use the computer account for High Performance Computing  purposes and not for commercial purposes. I will not use the computer account to engage in any form of illegal  software copying or other copyright infringement. I will not attempt to access accounts, files or information belonging  to other users without their knowledge and consent. I understand the monitoring of user activity on the Albuquerque  High Performance Computing Center machines may occur.

FAILURE TO COMPLY WITH THESE RULES WILL RESULT IN REMOVAL OF THIS ACCOUNT,  AND POSSIBLE DISCIPLINARY OR LEGAL ACTION.


Security
 Users are provided an initial Default Password in their new  user packet. You should change your password the first time you login  and periodically thereafter. If it is necessary for you to request a  password change from us (for example, when you have forgotten your current  password), your password will be set to this default. It is important  that you keep this form in a secure place for your reference. If you  do not have this form available, you will be required to wait until  a duplicate can be mailed to you via U.S. mail. AHPCC does NOT provide  passwords by phone, fax, or electronic mail.

General Considerations
The following considerations apply to running codes on any  AHPCC computer.

Users must regularly monitor their jobs to ensure that they  are not overburdening or hanging the machine they are using. This  is especially important for machines that are being used interactively.  Under UNIX, the ps, lsps, and vmstat commands provide  useful information about CPU utilization and the amount of paging  (memory utilization) that is going on in a system. Jobs running for long periods on machines that are used interactively  should be run with their priority reduced 10 units or more. (See the  man pages for the nice and renice commands.) Users should not lock the screen on terminals/computers intended  for public use when they are to be left unattended for more than a  few moments. Users should keep an Xterm window (console window) open to receive  system messages. All users should maintain current .plan, and if necessary, .forward files for their AHPCC accounts. These will allow us  to provide you with timely notification of system shutdowns, changes  to software, etc. They also allow us to contact you should a program  you are running cause a problem with the system. The .plan file should provide your name, current contact information (day and  evening phone numbers), and a brief description of the work you are  doing at the AHPCC. Your .plan file should also state whether  or not your AHPCC account is your primary account for reading E-mail.  If your AHPCC account is not your primary mail account, you should  also set up a .forward file to ensure that a copy of all your  messages are forwarded to your primary account. Access times. Access hours at the AHPCC are in general divided into  2 categories: prime time and non-prime time. During prime hours, which  run from 8am to 5pm on weekdays, priority access is given to workshops  or classes, students doing class work, parallel code development,  and production computation in that order. During non-prime hours,  which begin 15 minutes after the end of prime hours and end 15 minutes  before the start of prime hours, priority is generally given to production  computation, code development, and systems maintenance in that order.

Security Considerations
No passwords, for AHPCC or any other machines, whether inside or  outside the University, are to be kept in a personal file on the AHPCC  file system such as the .rhosts file, .netrc file, unless  the file is encrypted with strong encryption (e.g. DES or public  key encryption). Violation of this condition may lead to immediate  and permanent expulsion from AHPCC systems. Any such violation will  also be reported to the MHPCC and NCSA, since it is anticipated that  this condition will apply as well to any accounts eventually granted  on other HPCERC or PACI partner machines. For security reasons, it  is strongly recommended that users adhere to this policy on all network  machines on which he or she has accounts. For your own protection and that of the AHPCC community, the AHPCC  password shall consist of a minimum of 8 characters with no repeats.  At least one character should be a non-alphabetic character. The user's  name should not appear as any part of the password. Adherence to this  policy will be spot-checked with standard password-breaking security  programs. Users making insecure remote connections (e.g., telnet and rlogin)  or remote mail services should make it a practice to regularly change  their passwords. Once every three months is suggested. If possible  users should use secure connections, e.g. SSH (secure shell), etc.

Use of Silicon Graphics Visualization Computer  ("Nambe")
Nambe is a Silicon Graphics Onyx 2 purchased by the Visualization  Group at the AHPCC, and members of the Visualization Group therefore  have priority access to it. When not being used by the visualization  group, the primary use of the machine should be for running graphically  intensive applications, such as scientific visualization or computer-aided  design software.

Users should adhere to the following guidelines for using the machine  as a computational engine:

Production jobs, i.e. longer than 5 minutes, should be run with  their priority reduced by 10 or more. No single job should be submitted for a run exceeding 8 hours, and  if possible run at night. Runs of longer length should only be started  after consulting system administrators. Avoid using more than 4 processors simultaneously for serial or  parallel calculations. This will leave sufficient capacity for people  running most graphical applications on the machine. Parallel jobs using shared memory parallelization should be properly  tested and benchmarked before being run in a production mode. During prime hours, users should not run jobs requiring more than  50% of system memory.

Note: It is frequently necessary to reserve the machine for  demonstration or visualization group use, sometimes on short notice.

"Suitable notice" for a demonstration or other visualization  use is considered to be 24 hours. Notice should be given in the message  of the day (motd) at login. The hours for this use will be stated  clearly. If suitable notice is not possible, jobs may be terminated or suspended  without notice. Email notice will then be provided as to when suspended  jobs may be restarted.

Use of the Linux Clusters
The Alliance's Linux Superclusters, LosLobos and RoadRunner, are  for production. They operate as a cluster for parallel computing research  and development. Use of the Linux cluster for education compliments  its production computations.

The primary purpose of the AHPCC's Linux clusters are as an  educational and research tool. In the first role, they operate as terminals  for workshops and classes. In the second role, they operate as a cluster  for parallel computing research and development. Use of the Linux cluster  for production computation is a secondary purpose.

During regular working hours, and perhaps a few hours on either side,  users should not use the Linux cluster for computations. An exception  to this may be possible during scheduled university holidays, but only  with the consent of the system administrators.

Use of other workstations
Other workstations, e.g. IBM workstations, HP workstations,  and SGI workstations, are very limited in number and should mainly be  used for code development and for running applications that are specific  to those machines. Users should not use them for large or long computations,  particularly during the day. Computations run at night should be run  at very low priority so that dialup users may access the machines.

Use of servers
Systems that are designated as file servers, web servers,  and name servers should never be used for doing computations.  In general, users should avoid using them at all. The only exception  to the latter rule is when you are moving a large file to or from the  AHPCC file system, or between partitions on the AHPCC file system. In  this case, it is probably advantageous, because of the reduced network  traffic, to perform this operation directly on the file server. Consult  the AHPCC User's Guide on the AHPCC web site for information about servers.

© Copyright 1999-2001, Albuquerque High Performance Computing  Center