[daip] BUFSZ error

Eric Greisen egreisen at nrao.edu
Thu Jul 13 11:58:40 EDT 2000


David Kaplan writes:
 > Hi.  I was trying to run IMAGR on a large UV data set to produce 
 > wide-field faceted output.  However, just after it created the 25 MAP
 > files, it crashed with the following message:
 > quixot> IMAGR1: Beginning channel    1 through    1 with  2 IFs
 > quixot> IMAGR1: UVINIT: BUFSZ=    101376 TOO SMALL, LREC=65536

       Something has gone wrong here - LREC is very much too large.  I
am guessing that there are hidden values in the input UV data header
that are somehow being used to compute the LREC. 

 > quixot> IMAGR1: OUVOPN: ERROR   7 INITIALIZING UV DATA
 > quixot> IMAGR1: OUVOPN: ERROR OPENING UVdata work object
 > quixot> IMAGR1: IMACPY: ERROR COPYING Input UVdata
 > quixot> IMAGR1: TO UVdata work object
 > quixot> IMAGR1: Purports to die of UNNATURAL causes
 > quixot> IMAGR1: quixote      15APR99 TST: Cpu=       2.9  Real=       5
 > 
 > It looks like some sort of memory problem, but I can't find it mentioned
 > anywhere in the documentation.  I had been working successfully on this
 > data while at the AOC, using the machine OHSUMI (which has a lot of
 > memory).  Is this something I can fix here?
 > 

     No, machine memory has nothing to do with it.  The BUFSZ is our
standard BUFSZ and should hold any legal UV record (up to 16384
channels) but something has gone wrong.  What was done on this data
prior to running IMAGR?  Can you send me your inputs and place the
data set somewhere that I can fetch it?

     I note that you are running an AIPS that is more than a year old
- IMAGR has had several significant new options added plus a few bugs
fixed (not I think this one) in the interim.


Thanks, Eric Greisen



More information about the Daip mailing list