Daily Bulletin Archive

Nov. 29, 2011

Yellowstone Briefing
FL2 Large Auditorium (1022) or via Webcast
November 29, 2011 @ 2 p.m. MST

In concert with the Unidata Seminar Series, CISL presented a briefing on the Yellowstone system, NCAR's forthcoming data-intensive petascale environment, November 29, 2011, at 2 p.m. MST in the NCAR Foothills-2 Large Auditorium (1022) or via Webcast at  http://www.fin.ucar.edu/it/mms/fl-live.htm. The briefing discussed the Yellowstone hardware, including the disk resource and data analysis and visualization clusters, the software environment, and the allocations opportunities available to the various user communities.

The event was webcast and recorded for those unable to attend in person.

NCAR will continue to offer several allocations opportunities, the first of which will be for the Accelerated Scientific Discovery (ASD) period. A small number of large-scale, fast-turnaround projects in the geosciences will be selected for early, priority access from May-July 2012. The submission deadline for ASD projects will be January 13, 2012.

Yellowstone will be the inaugural computing resource at the NCAR-Wyoming Supercomputing Center (NWSC) in Cheyenne, Wyoming. Yellowstone will be a 1.6-petaflops IBM iDataPlex cluster with 74,592 processor cores and 149.2 TB of memory. The Yellowstone environment at NWSC will greatly expand the opportunities for researchers in the geosciences, with more than 600 million core-hours available each year.

For more details on Yellowstone, see:

Nov. 25, 2011

NCAR has announced that IBM will install a massive central file and data storage system, a petascale high-performance computational cluster, and a system for visualizing the data at the new NCAR-Wyoming Supercomputing Center (NWSC). Equipment delivery will commence in early 2012, with production computing operations planned for summer 2012.

The new system, named Yellowstone, and its data-centric supercomputing environment will be the NWSC's inaugural system. Yellowstone is expected to deliver 1.6 petaflops peak computing performance and provide nearly 30 times the computational performance of Bluefire. Yellowstone will be accompanied by a nearly 11-petabyte disk system.

For the full announcement, see the UCAR Communications site at:


For technical details, the CISL site has published initial information at:


Nov. 10, 2011

As part of NCAR's participation in the XSEDE collaboration (www.xsede.org), CISL is pleased to augment our training offerings with relevant courses offered by XSEDE partners. The following Fortran course from TACC is offered onsite in Austin, Texas, or can be viewed via webcast.

Fortran 90/95/2003 for HPC

November 10, 2011 (Thursday)
1 p.m. – 5 p.m. (CT)
J.J. Pickle Research Campus
ROC 1.603
10100 Burnet Rd.
Austin, TX 78758

Fortran is a modern language that is reviewed and updated regularly to meet the needs of the scientific community. It facilitates a traditional procedural programming style, but also supports object-oriented programming similar to C++.

Fortran programming skills are highly useful for developing new applications that achieve excellent performance, and for working with a large body of existing scientific codes that have been written in Fortran. These skills can be directly applied to using resources at computing centers and on any Linux/Unix/Windows environment. The class is intended for the intermediate user wishing to gain expertise in Fortran90/95/2003/2008 programming.

This class will be webcast.

To more information and to register, see https://www.xsede.org/web/xup/course-calendar. (Registration requires creation of an XSEDE portal account.)

Please submit any questions you may have via the XSEDE User Portal. https://portal.xsede.org/help-desk.

Nov. 1, 2011

CISL now offers the HPSS Tape Archiver (HTAR) utility, which lets users package files into a single archive file for efficient transfer to HPSS. With HTAR, there is no need to create and store archive files locally.

Documentation for this new utility is available here:


In addition to packaging files and transferring archive files to HPSS, HTAR creates and saves an index file in HPSS. It also allows retrieval of specified files from HPSS without copying the entire archive file to a local file system.

Oct. 20, 2011

Bulletin Update

Please note that the content of /blhome will become read-only for users after 1 p.m. on Tuesday, October 11. The content will remain that way for a month to enable users to copy the necessary files. After this period, /blhome will no longer be available.

Most users' home directories in Bluefire are now located under /glade/home. Directories that still appear under /blhome will be changed to /glade/home during the scheduled Bluefire outage on 10/11. If your home directory is under /blhome at present, you should already have received an email from us notifying you of this change.

The change of home directory means that the same dot files (.login, .profile, and so on) will be used on all CISL machines. If you wish to have settings that apply only to certain machines, you must set up your dot files to handle this. More instruction can be found at: http://www2.cisl.ucar.edu/docs/glade-user-guide/dotfiles .

If you have any questions, please contact us through cislhelp@ucar.edu or 303-497-2400.

Oct. 20, 2011

Recently CISL warned us about problems in older versions of NetCDF and attempted an update of the NetCDF libraries on bluefire. Because this change caused a disruption with currently running cases, and because the problem was rare and as far as we were aware we had never seen the issue, CSEG asked that the change be backed out and the older version be restored on bluefire. We have now found a case in which the old NetCDF file is causing a corruption of CESM output files and are thus recommending all CESM users make the following changes and recompile.

In file: scripts/ccsm_utils/Machines/Macros.bluefire replace the (-) lines with the (+) lines:

-NETCDF_PATH := /usr/local


-LDFLAGS := -q64 -bdatapsize:64K -bstackpsize:64K -btextpsize:32K

+LDFLAGS := -q64 -bdatapsize:64K -bstackpsize:64K -btextpsize:32K -lnetcdff

In file: scripts/ccsm_utils/Machines/env_machopts.bluefire

append the following two lines:

source /contrib/Modules/3.2.6/init/csh

module load netcdf/4.1.3_seq

This change should apply for all ccsm4 and newer versions. If you are using an older version please contact us. We apologize for the inconvenience, please contact Jim Edwards jedwards@ucar.edu if you have questions or concerns.

Oct. 20, 2011

Users are encouraged to use the /glade/scratch filesystem cross-mounted on Bluefire, Lynx and DAV clusters (mirage/storm). Each user should be able to store files under the directory:


The files stored in /glade/scratch will remain there for 30 days from the time of last access. There is no quota limitation, but the total space available in this filesystem is 204TB. Therefore, if you delete the files you do not need it will help your fellow users to utilize the space freed by you. For more details of usage and policy of this filesystem you may look at:


Apart from the convenience of not having any quota limitation, there are a number of other reasons to start using the /glade/scratch filesystem now. We mention a few of the important ones:

(1) The NWSC supercomputer, successor of our current computing systems, is scheduled to be commissioned around the middle of next year. There will not be any /ptmp filesystem on the NWSC system but we plan to deploy scratch filesystems similar to the current one though much bigger in size. Therefore, this is an opportunity to modify your scripts and workflows to make use of this scratch filesystem to make the transition to the NWSC system smoother.

(2) Due to some incompatibility in software configuration we will have to decommission the bluefire /ptmp filesystem before we bring up the NWSC system. During this time bluefire, lynx and DAV cluster will remain operational but bluefire's /ptmp will not be available.

(3) We want to monitor the behavior of this filesystem under high usage and also want to receive your feedback about usage of this filesystem. This information would allow us to fine tune our configurations and policies.

Please feel free to contact CISL Helpdesk at 303-497-2400 for any issues or concerns.

Oct. 17, 2011

On Saturday 15 October, 2011, the NCAR/CISL Data Center will be shut down to facilitate preventative maintenance on select infrastructure support systems. This is the normal Fall season down time for this purpose. Please note this activity is a complete shutdown with some minor exceptions listed below, scheduled for 6 a.m. to 6 p.m.

The supercomputer systems will begin their shut down at 12:01 a.m. Mountain Time while some computing systems, depending on their function, will begin their shutdown Friday afternoon/evening. Please make arrangements accordingly.

Once all maintenance work is completed, the systems will be brought back up in their order of dependency. CISL and Facilities Management and Sustainability maintain the twelve hour window to accomplish all tasks at hand.

Basic networking for phone service will be available for the Mesa Lab during most of this maintenance window, however, intermittent outages may occur.

For more information, contact CISL Help Desk at 303-497-2400

Oct. 17, 2011

Today, you will notice some changes to CISL's Daily Bulletin. We've modernized the underlying software, which allows us to provide you with important CISL information more conveniently and efficiently.

You'll still see a daily email each morning, but we can quickly send out urgent announcements when the need arises. In addition, on the Daily B's new web site at http://dailyb.cisl.ucar.edu (or via a convenient RSS feed), you can see items as soon as they are posted.  You can see events on the calendar as well as find past announcements in the archive.

We hope you find the new Daily Bulletin useful. If you have any comments or feedback, please send them to  dailyb@ucar.edu.

As always, you can change your Daily B preferences, subscribe, or unsubscribe online at http://mailman.ucar.edu/mailman/listinfo/dailyb-doc.