I believe this is purely because of the “Single Process” that the DevKit defaults to for running testing environments. What this does is every thing and, consequently every instance of every thing, is all stored within the same process all sharing the same memory, so the results will be inaccurate especially because everything is done differently - this may be a situation unique to ARK as EPIC have stated that results should be accurate and disabling the Single Process method should only need to be done if something is fairly broken(like the entire testing environment in our case haha).
I also believe that disabling the single process and running everything in standalone will provide much more accurate results, there is however a few caveats to this. One, it will take significantly longer to launch the test as it will run a dedicated server, and however many clients all in standalone processes(this would be a very bad idea when testing maps I believe - unless you’re running a workstation/development machine with the RAM and CPU power to handle such a thing) but the time it takes may not be worth it. Two, I’ve not gotten it working yet. I suspect that it’s mainly because all of the command arguments need to be specified but I mostly use ASM to run my dedicated so I’ve not learnt, nor do I even know where to look for, the arguments for running the servers.
I believe jslay is well aware of them, they are the issue however, as I’ve stated above.
-WM