Bug in how many cores to use with Optimizer?

[axel_kohonen]

[ FlexSim 16.2.0 ]

Hi,

I was running the optimizer in FlexSim 16.2 on my machine with 8 cores (intel i-7 4700MQ so hyperthreading) and noticed that if I run an experiment that only needs say 4 cores (e.g. 4 scenarios and 1 replication each) then if I run the optimizer after it then only 4 cores are used by the optimizer although 8 cores could be used as there are more than 4 different scenarios to consider for the optimizer. But if I run the experimenter so that it uses all 8 cores then the optimizer will use all the 8 cores if run after the experimenter. I am using the value of zero for "Max Cores" on the Advanced tab. Also, if I run the experimenter with 4 cores and then change the "Max Cores" value to 8 after the experiment is done then all 8 are used. But if I run the experiment again that uses only 4 cores then only the 4 cores are used for the next optimization round. If I stop the optimizer and run it again then all 8 cores are used. So the problem is seen only if running the optimizer directly after the experimenter when a lesser amount of cores was used in the experimenter than the optimizer could use.

It seems that the experimenter sets the value of used cores when run. Is this a bug or am I using the optimizer incorrectly? Or does it have something to do with the hyperthreading?

See the experimenter/optimizer tutorial model attached (experimenterexample.fsm) where the issue is seen.

Thank you!

Kind regards,

Axel

[philboboADSK · Accepted solution]

This has been fixed in 16.2.1.

[axel_kohonen]

Hi Phil,

Great! Thank you!

Axel