Hey Deko, why don’t you stoke cache to PRT and load back to PRT Loader then partition on second machine via deadline? Does that not work? It is not as direct but no less complex that the way we having been doing it for the last 6 years 
And yes some form of simple setup to partition would be cool 
Stoke is very well multi-threaded, so running multiple Stoke sims on the same machine is out of the question.
The simulation itself is currently run in a MAXScript loop inside the UI handler, but we are working on splitting that to a separate function, so sending the same scene to Deadline and running a simple post-load script to change the Random Seed, the Path and run the simulation as a MAXScript job should be relatively easy to do once we have that sorted out…
Right now you can run the same scene on multiple machines (or sequentially on the same machine) with a different random seed and it should produce the variations you need. Just change the prefix name from “Stoke” to “Stoke_part01of10_” and so on…
That was I already do. 23 partitions in 2 days will not my wish for the next test 
But it works. Results are great as expected.
Will look in to simple script after the weekend.
A little offtopic:
Bobo, can you a little more explain what is ‘Vary Randomness Over Time’ do?
Thank you.
I figured as much but had to ask
You do use concurrent tasks too right? Meaning you are saturating all of yours cores with each single threaded partition assignment.
I haven’t tried but I can’t imagine it would be that much faster using Stoke to generate those partition after the first pass it done. Maybe I am wrong, Bobo?
Nope, as Bobo mentioned, Stoke use much of my workstation (about 50-70% of 24 cores). So I used old good method - one by one partition, manually