[Date Prev][Date Next][Thread Prev][Thread Next][Date Index][Thread Index]

Grand Challenge queues on mcurie



Dear NERSC User,

Two queues have been set up on mcurie to handle jobs for Grand 
Challenge projects. The queues can run jobs between 128 and 256 PEs 
with a duration of between 2.5 to 12 hours. You should use the QSUB 
parameters mpp_p and  mpp_t to indicate your requirements for number 
of PEs and time respectively. NQS will automatically route the job to 
the correct queue. These queues are named gc128 and gc256.

Initially, there will be a limit of one running job per user in these 
queues. Jobs submitted to either of these queues will count towards 
the total of 3 jobs you are allowed in the NQS batch system. 
For example, if you submit two jobs to the gc queues, you will be 
allowed only one job in the normal queues.

These queues will run during the evening and night for 8 hours per day.
If your job does not complete during a shift, it will be checkpointed 
and then restarted during the next period. Remember that the queue 
limit is 12 hours, you are responsible for checkpointing your code if 
it requires more time than this. The normal queues will continue to be
available during the day shift, and you can continue to make use of 
these.

If your job gets checkpointed, it will not be visible using the ps, 
grmview or tstat commands. Instead, it will be listed in NQS as held:

mcurie$ qstat -a
---------------------------------
NQS 3.2.1.5 BATCH REQUEST SUMMARY
---------------------------------
IDENTIFIER    NAME   USER    LOCATION/QUEUE  JID  PRTY REQMEM REQTIM ST
------------  -----  -----   --------------  ---  ---- ------ ------ ---
2763.mcurie   job3   demo    pe64@mcurie           ---   4096     60 Hop

Please contact NERSC by email at consult@nersc.gov, or by phone at
1-800-66-NERSC, if you have any comments or questions.

Jonathan Carter
NERSC User Services

------------- End Forwarded Message -------------