I came across the a weird effect:
When I run the optimizer with just one configuration, the simulation runs much faster (twice the speed), as when I run the same configuration several times (e.g. by optimizing on a bogus parameter).
This is especially evident when I'm running more configurations than cores.
I'm running of a 4-core i5 with 8GB RAM.
So:
- Any idea why this happens?
- Is there some shared resource the simulation threads are trying to use simultaneously?
- Can I limit the number of worker threads spawned by the simulator?
Thanks,
Boaz
Comment