Skip to main content

number of threads in solution pool method

Answered

Comments

3 comments

  • Matthias Miltenberger
    Gurobi Staff Gurobi Staff

    Hi Akram,

    There is no guarantee that Gurobi is going to utilize all available cores at all times during the solving process. It is fairly common that especially during the beginning and the end, there simply is not enough work to distribute among the cores/threads.

    Cheers,
    Matthias

    0
  • Akram Kh
    Gurobi-versary
    Conversationalist
    Curious

    Dear Matthias,

    Thanks for your reply. Actually I have a big model which takes some days to be solved. I also have worked with gurobi to solve such big models without using solution pool  and monitored its resource consumption (RAM and CPU). Although, as you have noted, CPU is not fully utilized at the beginning and the end of process, but in the middle it is(not to say all the cores but most of them are engaged).

    The problem is that when I try to solve my model using solution pool, finding the first optimum solution as well as the first part of solution pool completion utilizes most of the cores, however it switches to single thread mode very soon (e.g. after  4 hours in a model which takes 10 days to be solved including solution pool completion).

    Do you have any idea about this observation?

    0
  • Jaromił Najman
    Gurobi Staff Gurobi Staff

    This post is continued in Number of threads in solution pool method.

    0

Please sign in to leave a comment.