ARSC T3E Users' Newsletter 162, February 19, 1999

Yukon Update Successful

Below is a fairly typical "grmap" snapshot of yukon from this week... Thanks and congratulations to the SGI teams from Chippewa falls and Fairbanks who upgraded yukon last weekend.


     UserName Size BasePE      Mem avg:max  Command       
     ======== ==== =========== ============ ==============
 A -   abcde  60   0   [0x0  ]      90:93   a.out                         
 B -   bcdef  14   60  [0x3c ]     102:103  b.out                         
 C -   cdefg  72   100 [0x64 ]      71:84   c.out                         
 D -   defgh  72   172 [0xac ]     188:188  d.out                         
--------------------------------------------------------
 0  
 <A..A..A..A..A..A..A..A. .A..A..A..A..A..A..A..A.
 16 
 .A..A..A..A..A..A..A..A. .A..A..A..A..A..A..A..A.
 32 
 .A..A..A..A..A..A..A..A. .A..A..A..A..A..A..A..A.
 48 
 .A..A..A..A..A..A..A..A. .A..A..A..A><B..B..B..B.
 64 
 .B..B..B..B..B..B..B..B. .B..B>< .. .. .. .. .. .
 80 
 . .. .. .. .. .. .. .. . . .. .. .. .. .. .. .. .
 96 
 . .. .. .. ><C..C..C..C. .C..C..C..C..C..C..C..C.
 112
 .C..C..C..C..C..C..C..C. .C..C..C..C..C..C..C..C.
 128
 .C..C..C..C..C..C..C..C. .C..C..C..C..C..C..C..C.
 144
 .C..C..C..C..C..C..C..C. .C..C..C..C..C..C..C..C.
 160
 .C..C..C..C..C..C..C..C. .C..C..C..C><D..D..D..D.
 176
 .D..D..D..D..D..D..D..D. .D..D..D..D..D..D..D..D.
 192
 .D..D..D..D..D..D..D..D. .D..D..D..D..D..D..D..D.
 208
 .D..D..D..D..D..D..D..D. .D..D..D..D..D..D..D..D.
 224
 .D..D..D..D..D..D..D..D. .D..D..D..D..D..D..D..D.
 240
 .D..D..D..D>< .. .. .. . . .. .. .. .. .. .. .. .
 256
 . .. .. .. .
--------------------------------------------------------
APP-PEs tot:used:free:down  MEM tot:used:free  JOBS run:blk
        260: 218:42  :0          63:  26:37           4:0   

As can be seen, the upgrade has already increased the ARSC capacity by letting more jobs run at the same time, helping to reduce waiting times for results. Some users have already used the entire system for higher resolution models.

(For a copy of the grmap script, see newsletter 152. /arsc/support/news/t3enews/t3enews152/index.xml )

New Yukon Queues

From: "news queues" on yukon:



   T3E Batch Queues

   ================

   The queue configuration (as of February 15, 1999) is:



   Name of        Number    Maximum jobs   Time limit    When queue is

   queue          of PEs     per queue      on queue      executable

   -----------  ---------  ------------   ----------   ----------------

   grand        161 - 256       1          4  hours        always *

   xxlarge      101 - 160       1          4  hours        always

   xlarge        51 - 100       1          4  hours        always

   large         21 - 50        3          8  hours        always

   medium        11 - 20        4          8  hours        always

   small          2 - 10        4          8  hours        always



   Qgrand       161 - 256       1         30 minutes       always *

   Qxxlarge     101 - 160       1         30 minutes       always

   Qxlarge       51 - 100       1         30 minutes       always

   Qlarge        21 - 50        3         30 minutes       always

   Qmedium       11 - 20        4         30 minutes       always

   Qsmall         2 - 10        4         30 minutes       always



   single              1       10          8  hours        always





   Please use the minimum number of PEs required to obtain sufficient

   memory for your job to run (memory per PE = 256 MBytes), this will

   improve system throughput for ALL users.





   (*) The processing of jobs in the grand queue (>160 PEs) is

       dependent on system load. In general, smaller jobs are more

       likely to run than larger jobs.   Contact ARSC consulting

       (consult@arsc.edu) to discuss special arrangements for the

       running of extremely large jobs.  See: "news chkpnt_sched"

       for the current schedule on which such jobs may run.  See:

       "news holding_jobs" for general information on the check-

       pointing of jobs.


User feedback is important as we decide whether to leave the queue structure as is or to make adjustment. Please let us know if it's not working for you, or if it is.

Three Conference Announcements

CUG '99

41st Cray User Group Conference May 24-28, 1999 Go to http://www.cug.org and register today! Minneapolis, Minnesota

As the 20th century comes to a close and we look forward to the 21st, the CUG conference returns to the birthplace of supercomputing.

The Program Committee for the Spring 1999 CUG Conference in Minneapolis is planning a top-notch technical program for users and administrators of all high-end SGI platforms. Several SGI general sessions will focus on high-end product progress and high-performance computing strategies for the future.

P A R C O  '99

PARALLEL COMPUTING '99 17 - 20 Augustus 1999 Delft, The Netherlands WWW: http://www.parco99.tudelft.nl

ParCo99 is the continuation of a series of international conferences on parallel computing started in 1983. The collected proceedings give an overview of the achievements in the Parallel Computing field over the last two decades of the 20th century. On the threshold to the next century ParCo99 aims to give a retrospective and a prospective view of parallel computing.

PARALLEL CFD '99

May 23-26, 1999 Williamsburg, Virginia, USA http://www.parcfd.org

Parallel CFD'99, the eleventh in an international series of meetings featuring computational fluid dynamics research on parallel computers, will be held May 23-26, 1999 in Williamsburg, Virginia.

Quick-Tip Q & A


A:{{ Which is better at retrieving multiple migrated files?
        yukon%  dmget A B C
          <or>
        yukon%  dmget A
  }}

  In general, the first would be preferred.  Doing it that way, DMF can
  batch the recalls into one task if possible.  That is important
  because it means DMF can get all three files with one load and read
  of a single tape.  The second method could mean the same tape gets
  loaded and read through three times.

  From the dmget manpage:

       To efficiently recall all files in a directory named mydir and
       its subdirectories, use the following command:

           find mydir -type m -print 
 dmget

  NOTE - only one execution of dmget and the file list is potentially
  huge.



Q: You want extract clusters of lines from a file based on a start and
   end condition.  For example, you might want to print all comment
   lines from a C source file (and no other lines).  Here's a rough
   algorithm:

    1) search for "\*"  (the start of a C comment)
    2) print that line.
    3) search for the next "*\" (the end of the  C comment); printing 
       every line along the way.
    4) print the line containing "*\".
    5) stop printing lines.
    6) GOTO step 1)
  Can you give a Unix command to do this?

[ Answers, questions, and tips graciously accepted. ]


Current Editors:
Ed Kornkven ARSC HPC Specialist ph: 907-450-8669
Kate Hedstrom ARSC Oceanographic Specialist ph: 907-450-8678
Arctic Region Supercomputing Center
University of Alaska Fairbanks
PO Box 756020
Fairbanks AK 99775-6020
E-mail Subscriptions: Archives:
    Back issues of the ASCII e-mail edition of the ARSC T3D/T3E/HPC Users' Newsletter are available by request. Please contact the editors.
Back to Top