Is there any way to estimate the additional processing time required for concurrent runs of ConQuest?

Question

It's not uncommon for me to run multiple scaling jobs in ConQuest [IRT software]. I'm curious if there is any way to estimate the amount of additional time required when running concurrent sessions. Obviously this is dictated by available resources (e.g., CPU speed and utilisation) and I imagine the number of parameter being estimated, but it would be good to know if there was a rule of thumb that can be applied. For example, more than four concurrent jobs will slow things down to the point that it's not worth doing.

I commonly run two ConQuest jobs at a time and haven't noticed much degradation in time to completion.

Answer 1

Without doing any benchmarking, I would suspect the answer is that as many concurrent instances of ConQuest as you have cores should give you linear efficacy.

That said, I would optimise other aspects first - minimise disk IO via command set to speed up runs.

As always, YMMV.