Just some random facts then... :)
My system is Win 10 x64 1803. Hardware-wise it's Intel 6-core Core i5 9600K with 32GB RAM and Samsung SSDs for storage. So, I would imagine it's a reasonably powerful desktop.
As for the setup, yes I am kind of a power user I suppose (though I almost exclusively use Tree view, so in that regard not a power user per se). But I do run 6 TDL instances at a time, each of which have around 5 to 10 task list files open, each with anywhere between 20 to 1000 tasks.
While it may seem like a lot at first, on a per instance, and especially on a per task list basis, it's not that much. I mean even if you consider 1000+ items, it should not be too demanding to perform all the basic operations without hurting responsiveness.
And yet, even switching between task lists takes on average 10 seconds, even when the task list has been loaded. Opening all the task lists (within an instance of a program, when there are around 10 of them) is on the order of 30 seconds, even with delayed load.
Task entry (create new task, change attributes) is ok, but certainly not smooth or great. There is a noticeable delay in the UI that is easily visible between when you type something or create new task and it shows up - usually on the order of a second or sometimes even more.
This behavior seems to be the case regardless of how and if the tasks are sorted or filtered, although the latter certainly introduces even more slowness.
At the same time I notice how excellent TDL is with RAM usage - it uses very little of it, even with largeish task lists and with having 10 of them open. On the order of 50MB for Private Bytes - nothing in comparison to say Chrome, where a single tab can require more RAM. So, perhaps the program can be sped up considerably by aggressive caching, i.e. use more RAM and less CPU where appropriate. I mean, I would not mind if TDL started using as much as 4 or even 6 times more RAM if it would work smoothly.