Skip to main content

Variation in runtimes for big linear problem

Victor Hugo Nascimento
  • Gurobi-versary
  • First Question

I'm solving a big continuous problem that is presenting high variations in the run times. For example, I've run the program 3 times and got the run times: 175.86, 341.24 and 190.98 seconds. According to the logs, the barrier method was used. Is this variation in run times normal? I know there are OS issues (my OS is Linux Ubuntu 16.04) that may cause variations, but other sections of my program, like the time to load the model and write the solution to a file, have small variations in run times (60.87, 66.29 and 60.75 seconds respectively.).

1

Comments

2 comments

Date Votes
  • Official comment
    Simranjit Kaur
    • Gurobi Staff
    Edited
    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?.
  • Silke Horn
    • Gurobi Staff

    When solving the exactly same model multiple times on the same machine, the running times should be similar.

    • Are there other processes running on the same machine at the same time that may slow down Gurobi?
    • Is it possible that your code creates slightly different models each time you run it? E.g. a change in the order of the constraints can lead to a different solver path and could easily explain the variantions you are seeing. To find out, you could write the model to an MPS file for multiple runs and then compare the files. Can you reproduce this behavior when solving from an MPS file?

     

    0

Post is closed for comments.

Didn't find what you were looking for?

New post