- 2 GB RAM is desired and preferred (though for the AXAF data 1 GB will do if one 
  uses ~ 3000 x 3000 pixels for the main field).

- Please make sure that in the file '.aipsrc' in your home directory
  there is the following line that defines how much memory the system
  has (e.g if your machine has 2Gbyte) :

  system.resources.memory: 2000


- 32 bit machines can handle only 2 Gbyte per process (independent of
  the amount of memory, real and swap, the machine has )

- Using 7200 x 7200 pixels image without facetting makes imager come
  very close to the limits of 2 Gbytes (and it crosses it at times:
  see next point)

- When running a process near the amount that can be addressed may
  give non repeatable crashes, at times it may run through or at other
  times it may crash. This is dependent on the state of fragmentation
  the memory available is. E.g when trying to access a subset of an
  array, when the memory is not fragmented, is done without the need
  to copy. But if the data is dispersed, then a copy is necessary and
  may lead to a crash as the 2 Gbyte limit may be crossed in this
  copy.

- If imager crashes with 'st9bad_alloc' message try to change things so
  that it uses less memory (one or all of the following will help):

  (a) reduce image size.
  (b) Use facets, effectively reducing image
  size used at a time. Also please note that wprojplanes can be less
  if more facets are used.
  (c) try to use Hogbom clean instead of Clark
  clean (i.e use mfhogbom or wfhogbom instead of mfclark or wfclark).

  (b) and (c)  above steps sacrifice computation speed for less memory usage 

- If glish or the logger becomes unresponsive but imager is running
  (as seen in 'top' say) then one can safely let imager run to
  completion. One may even CRTL-C out of glish, imager will continue
  to run unless explicitely killed.

- Patience is needed: 1046 data, reduced by the example script took
  ~24 hours on basho. Using flanking fields or more facets will
  increase the time of reduction. (Same script but with 2 x 2 facets
  took nearly 48 hours on atlas)