Serial cluster throughput increase

[LRZ aktuell]

A change in the scheduling policies has been introduced to increase the
 overall throughput.
 
 We slightly decreased the default setting for memory from 2000 MB to
 1990 MB. This will help to schedule more jobs to a node, but should not
 affect users severely.
 
 Users who have previously submitted jobs do not need to change their
 pending jobs in the queue.
 
 Newly submitted jobs will automatically take advantage of the enhanced
 throughput. (Some users might want to cancel their pending jobs and
 resubmit them to take advantage of it)


 This information is also available on our web server
 http://www.lrz-muenchen.de/services/compute/aktuell/ali4625/

 JuanPancorbo Armada