TobiasAchterberg 270002SFM98 Posts
Re: solving multiple small problems vs. one large problem2013-09-10T15:15:48ZThis is the accepted answer. This is the accepted answer.
If this are only LPs (linear constraints, linear objective, only continuous variables), then solving them as a single big LP should usually be roughly similar in performance than solving them individually. It is a question of overhead for calling the LP solver versus the overhead of having to deal with larger memory footprints and larger basis matrices.
If those are mostly easy problems, then you should try to solve them individually and only use a single thread (set the "threads" parameter to 1). This should already eliminate a great portion of the set-up overhead in the LP solves. Moreover, you should check whether the barrier algorithm is faster than the dual simplex.