Skip to main content

Overloading Compute server ?

Answered
Kristy Naylor
Gurobi-versary
First Comment
First Question

Hi there, I am wondering what the repercussions of overloading the compute server with jobs is? i.e. sending 3 jobs to the compute server which all used 8 of 8 available threads/cores?

My understanding is this may impact speed (?) but will this impact on the resultant solution?

I have a complex assignment problem which was solving nicely and as expected, i.e. closest geographical assignment but with new set of data the assignment is not as expected, and the one difference (apart from data) is that I ran the job when other problems were solving on the compute server.

thanks for your help,

Kristy

0

Comments

2 comments

Date Votes
  • Official comment
    Simranjit Kaur
    Gurobi Staff Gurobi Staff
    This post is more than three years old. Some information may not be up to date. For current information, please check the Gurobi Documentation or Knowledge Base. If you need more help, please create a new post in the community forum. Or why not try our AI Gurobot?.
  • Matthias Miltenberger
    Gurobi Staff Gurobi Staff

    Hi Kristy,

    I hope this reply is still relevant - please excuse the long delay!

    Sending multiple jobs to the Compute Server should not change the solver's path to optimality (given that no parameters have been changed). Every single job on the server will try to use all available cores and generate as many threads accordingly. Overloading the server usually leads to worse solving times because the OS on the server needs to juggle all those threads. That's why the default JOBLIMIT for parallel jobs is set to 2 to avoid this congestion and rather make use of the queueing mechanism of the Compute Server.

    Cheers,
    Matthias

    0

Post is closed for comments.

Didn't find what you were looking for?

New post