Solving multiple CPLEX models in parallel

Question

I have some docplex models that I need to populate solution pools for them at the same time. All the models have a lazy constraint callback. My problem is when I start solving these models at the same time ,by running them on different consoles, their runtime increases. 1 single model can populate in 200 seconds but when I start solving 3 different models at the same time the runtime for that model becomes 2000 seconds. assuming that I have enough CPU and memory, why is this happening? and how can I avoid it and get the lower runtime?

Answer 1

You could try to limit each solve to 1 thread by using threads.

As an example

from docplex.mp.model import Model

mdl = Model(name='buses')

mdl.parameters.threads=1


nbbus40 = mdl.integer_var(name='nbBus40')
nbbus30 = mdl.integer_var(name='nbBus30')
mdl.add_constraint(nbbus40*40 + nbbus30*30 >= 300, 'kids')
mdl.minimize(nbbus40*500 + nbbus30*400)

mdl.solve()



print("threads = ",mdl.parameters.threads.get())


for v in mdl.iter_integer_vars():
    print(v," = ",v.solution_value)

where I changed

https://github.com/AlexFleischerParis/zoodocplex/blob/master/zoosettings.py

to show how to set the treads parameter