I would like to use GNU Parallel as a queue system/batch manager. This will be especially useful with the "-S" feature. When I start it up (as in the example in the man page), I submit JobSlots number of dummy jobs to initially fill the queue. Up until now that number has been easy to know, being on the local machine. But how can I know how many JobSlots there are to fill before subsequent commands start executing? Doesn't it depend on how many CPUs or cores or threads Parallel finds on the remote machines (which could depend on which Parallel profile is selected)? Is there a way to find out, maybe with a test command in advance of the actual run?
-- Thanks, Joe
